robotic system engineers

Abstract: Reliable box manipulation requires successful placement without damage, which depends on maintaining sufficient contact force without excessive loading. We hypothesize that per-finger force feedback improves manipulation reliability relative to open-loop finger commands. To this end, we present Finger External-Force Estimation from Proprioceptive SignaLs (FEEL), an uncertainty-aware residual method that estimates per-finger external force magnitude from proprioceptive histories. Our method combines a calibrated virtual-spring estimate with a residual distribution predicted by a Mixture Density Network (MDN) and requires no dedicated contact sensors during deployment. The estimated force is used for continuous per-finger adaptation, while predictive uncertainty is evaluated as an empirical signal of ambiguity in proprioceptive force estimation. Across ten fingers and contact forces spanning 0-20 N, our method achieves a mean RMSE of 0.251 N, reducing the error by 92.8% relative to the virtual spring and by 9.4% relative to an end-to-end MDN. In box pick-and-place, force adaptation maintains task success at 10/10 and increases both damage-free outcomes and reliability from 7/10 to 10/10. The project page is available at https://finger-force-estimation.pages.dev/

Abstract: Deep reinforcement learning (RL) algorithms have been observed to lose the ability to adaptively learn from new data, a phenomenon termed Loss of Plasticity (LoP). While many metrics have been shown to correlate with it, the causal mechanisms and the geometric conditions from which it can arise have not been clearly identified. We analyze LoP in Proximal Policy Optimization with isotropic Gaussian policies, the state-of-the-art formulation for many continuous action spaces in RL. We report multiple metrics which characterize the effect that the policy standard deviation $σ$ has on LoP, and we also suggest related explanations for the effectiveness of widely adopted LoP mitigation techniques. We observe a significant effect of $σ$ on the extent of LoP, and accompany this with both theoretical and experimental supporting arguments. Our experimental validation consists of LoP benchmark evaluations against standard baselines as well as real-world policy deployment on a 7-DoF robotic arm.

Abstract: Humans efficiently process multimodal streams spanning hours to years, building persistent, fine-grained spatiotemporal memories that support reasoning about and predicting long-horizon events. In contrast, existing benchmarks focus on short- and medium-context tasks, and most foundation models struggle to preserve fine-grained information over long horizons. To evaluate and improve long-horizon spatiotemporal reasoning, we introduce the LongVideo dataset and the 4DMemoryBench. LongVideo comprises about 900 hours of video and over 5,000 manually reviewed questions covering visual, textual, and spatiotemporal reasoning across diverse context lengths. We further develop the 4DMemory Harness, which integrates four complementary skills, 22+ specialized tool models, and a model-specific retrieval-augmented generation system. Evaluating 13 leading models, we find that standalone foundation models struggle with long-horizon tasks, while the 4DMemory Harness improves all of them by over 50\%. Compared to previous benchmark, the same models exhibit a performance drop of more than 10\% on 4DMemoryBench. We hope 4DMemory will facilitate the development of spatiotemporal agents capable of addressing complex, long-horizon tasks.

Abstract: Physical self-reproduction in modular robots remains difficult because a robot must not only move, but also collect, align, dock, and separate modules reliably in hardware. Taking inspiration from how genetically identical cells form the varied tissues of an organism, we introduce a single standardized robotic module that combines a magnetic docking interface with two actuated joints and folds from a flat, two-dimensional configuration into three-dimensional structures. Modules of this single design can be assembled into two morphologies: a three-module "trilobite" and a five-module "seal". Using a small set of discrete motion commands, a human operator steers an assembled parent robot to dock magnetically with a sequence of spare modules, which are assembled into a second, structurally identical individual that then separates from the parent. Because the parent retains all of its original modules, this is a budding-style form of physical reproduction. Here, we report preliminary hardware demonstrations of this full process on both morphologies. We discuss the current scope of the system, including its present reliance on direct human teleoperation rather than autonomous control, and outline directions toward autonomous execution of this process and repeated validation across a larger number of independent trials.

Abstract: Dynamic representations based on 4D Gaussian Splatting (4DGS) have shown strong capabilities in reconstruction and rendering, yet current methods remain limited in supporting complex natural-language referring in dynamic scenes. Existing 4D Gaussian methods are primarily designed for rendering-oriented dynamic modeling, while current semantic extensions often rely on scene-specific semantic optimization or are evaluated on relatively simple query settings. In this paper, we study Referring 4D Gaussian Splatting (R4DGS), a task for grounding temporally varying, multi-target or reasoning-intensive, and zero-target or distractor expressions in dynamic 4D scenes. To support this task, we introduce R4D-Bench-QA, a benchmark with structured query annotations. We further present ReferGaussian, a unified framework built upon 4DGS that converts low-level dynamic Gaussians into persistent, language-addressable entities for training-free referring inference. ReferGaussian organizes mask-supported Gaussian entities in an EntityBank, uses Multimodal Large Language Models (MLLMs) to decompose complex queries into trackable object phrases and structured referring constraints, and performs constraint-aware grounding without scene-specific semantic optimization. Experiments on R4D-Bench-QA and the 4D LangSplat protocol demonstrate improved temporal target selection and competitive spatiotemporal grounding without scene-specific semantic training. Project resources are available at https://chenbangpu.github.io/R4DGS/.

Abstract: Robots exploring unknown indoor environments must decide where to move while much of the environment remains unseen. Predicting the occupancy of unseen space beyond the current sensor range can improve these decisions, but recent approaches commonly rely on learned models that require offline training and repeated inference during exploration. We introduce the Mondrian-Anchored Prior for Learning-free Indoor Exploration (MAPLE), a training-free predictive approach that adapts the Mondrian process into an online generative prior conditioned on observed geometric cues in the current partial map. MAPLE generates multiple plausible map hypotheses and evaluates visibility separately within each to estimate frontier gain, preserving differences in predicted visibility across hypotheses. Across KTH floor-plans and real-SLAM environments at multiple sensor ranges, MAPLE consistently improves over non-predictive exploration and remains competitive with learned predictive approaches. Its lightweight procedural construction enables predictive exploration with limited computational resources. Experimental results further show that more accurate map prediction does not necessarily produce better exploration, highlighting the importance of how predictions are represented and used in exploration decisions.

Abstract: The integration of dynamic legged locomotion and aerial mobility offers versatility that is deterred by challenges in payload distribution, actuation, and the contrasting requirements of both domains. In this paper, we present the design and operation of a multimodal quadrupedal aerial-legged robot. The presence of legged aerial systems also allows for thruster-assisted legged locomotion, which we make possible with a unified control framework using Control Contraction Metrics (CCM). We search for separate domain-specific contraction metrics for walking and flight independently offline and unify them using an online energy-based Control Lyapunov Function Quadratic Program (CLF QP). This enables optimal, cross-domain control allocation, which we show using simulations. We also show experimental results that validate the mechanical viability of our hardware and the paradigm of thruster-assisted trotting. The stability of said experiments motivates the deployment of the control framework for multimodal operations.

Abstract: Retrieval tasks are a form of search task where a robot must explore, find an object, and return with it as quickly as possible. Recent work in ergodic control has demonstrated time-optimal path planning for search tasks where visitation is proportionally matched, within some tolerance, to a given target distribution. However, these planners minimize the duration of a full deterministic search, while retrieval missions end their search at an unknown point in the trajectory and must return to a given base. In this work, we propose a new formulation of the time-optimal ergodic search problem that models detection as a Poisson process and minimizes total mission time by including the return phase in the objective. Experimental results show that time-optimal ergodic retrieval (TOER) achieves up to $19\%$ shorter mission durations relative to ergodic baselines and that this advantage grows as the return phase becomes more costly. Simulation and hardware implementations validate TOER's compatibility with environments of non-convex geometry as well as both 2D and 3D search domains.

Abstract: Robotic manipulation is inherently geometric, but what shape does the world take inside the policy that performs it? Prior work on vision-language-action (VLA) policies establishes that task-relevant variables are linearly readable from activations, yet readability certifies only that a variable is present and says nothing about the shape its values trace, which may be curved and multi-dimensional. Using controlled sweeps, natural rollouts, linear probes, and activation interventions, we build a taxonomy of seven task-relevant concepts in VLA models. We find that realized geometry is predominantly multi-dimensional and is shaped by the constraints under which the policy acts. Physical joint limits, for example, leave wrist orientation an open arc rather than a circle, and distance to the target, a single scalar, is realized in two orthogonal subspaces, a coarse ruler for the approach and a fine ruler for the grasp. The same geometries reappear in two further VLA models on real-world robot data, indicating that they are neither model-specific nor artifacts of simulation. We demonstrate how the findings are actionable by steering along the geometry a concept realizes. Overall, this work lays out an initial phenomenology of realized geometry in VLA policies and demonstrates that a concept's shape carries as much importance as its readability.

Abstract: \emph{Dispersion} requires a set of mobile agents to occupy distinct nodes of a graph such that each node contains at most one agent. We study deterministic dispersion of $k\le n$ agents on an $n$-node oriented square grid, where each node has globally consistent North, East, South, and West port labels, from arbitrary initial configurations. This includes both \emph{rooted} configurations, where all agents are co-located initially, and \emph{general} configurations, where agents are initially distributed arbitrarily across the nodes. On oriented square grids, Kshemkalyani, Molla, and Sharma~\cite{KMS20} gave an $O(\min(k,\Side))$-round algorithm in the local-communication model, where only agents on the same node can exchange information, and an $O(\sqrt{k})$-round algorithm in the global-communication model, where any two agents can communicate regardless of position, both using $O(\log k)$ bits per agent. The bound in the local communication model is optimal when $k=Θ(n)$, but it remains open whether local communication alone can match the $Ω(\sqrt{k})$ time lower bound when $k=o(n)$. We solve this by presenting \textsc{Spread--Reverse--Collect} (SRC), a deterministic algorithm that disperses any $k\le n$ agents in $O(\sqrt{k})$ synchronous rounds using $O(\log k)$ bits of persistent memory per agent, matching both time and memory lower bounds. SRC is based on a geometric primitive, a \sharpS-shaped parallel sweep, together with a simple priority rule. Unlike earlier grid algorithms that gather agents at the boundary and spread out from there, each SRC sweep starts from the agents' current position, and when it reaches the boundary the group simply reflects its sweep back inside and continues. SRC therefore needs no knowledge of the grid size and runs unchanged on an infinite oriented grid.

Abstract: Diffusion-based policies have recently emerged as a powerful paradigm for offline reinforcement learning, but their reliance on multi-step denoising introduces substantial computational overhead and requires backpropagation through time during training, limiting scalability and practical deployment. In this work, we propose \textbf{Instant Flow Q-Learning (InsQL)}, a single-step generative policy learning framework that retains the expressive capacity of diffusion models while eliminating iterative sampling and BPTT. InsQL builds on the flow-matching perspective of diffusion policies, but replaces marginal velocity parameterization with \emph{average-velocity} modeling and introduces a principled compositional mechanism, yielding a theoretically grounded objective that learns average velocity consistently across scales. This formulation provides a direct noise-to-action mapping and enables single-step action generation. InsQL further unifies compositional average-velocity modeling with behavior-regularized actor--critic optimization in a single framework, requiring only a single policy network and avoiding distillation, multi-stage procedures, and Jacobian computations during training. At inference time, actions are generated in a single forward pass and can be refined via value-guided candidate selection to improve performance. Extensive experiments on standard offline RL benchmarks and real-world robotic manipulation tasks show that InsQL matches or surpasses state-of-the-art diffusion-based methods while substantially reducing both training and inference time. Overall, these results establish compositional average-velocity modeling as an efficient and scalable alternative to multi-step generative policies for offline RL.

Abstract: Robust motion planning and control for manipulators is the problem of guaranteeing safe execution and collision-free motion under disturbances. The Kinodynamic Motion Planning via Funnel Control (KDF) is a recently proposed planner that comes with robustness guarantees and is particularly attractive for collision-free control as it guarantees closed-loop evolution inside a funnel, i.e., a collision-free region around a nominal trajectory. Despite its attractive theoretical properties, the current approach for computing funnels is computationally demanding, which limits its expressiveness and performance. We address its current limitations by proposing a novel funnel planning and control solution. Our key tool to enable this is a neural Signed Configuration Distance Function (nSCDF) which allows us to estimate collision-free balls and gradients in a computationally efficient way. We use the nSCDF in our planner to compute a nominal trajectory, where path length and clearance are balanced. To track the trajectory with collision-avoidance guarantees, we design a funnel controller that exploits the geometric information from the nSCDF. From numerical experiments, we observe that our planner produces nominal paths $2$ times faster, while at the same time obtaining $20 \%$ shorter path length and $5$ times the clearance. Finally, we also observe improved motion performance and robustness properties when using the nSCDF for funnel control.

Abstract: Camera-first agents must act on the visual world as it unfolds. We argue that the central obstacle for future camera-first agents is not recognition ability - multimodal models (vision-language and live/streaming alike) generalize well through language and remain the right tool for semantic reasoning - but time. Many visual targets are ephemeral: they appear, are query-relevant, and leave within a short window. Binding memory to the same end-to-end path as reasoning means that by the time a monolithic pipeline completes, the evidence is gone. We formalize this as Ephemeral-Target Interaction (ETI) and a Perceptual Deadline Contract separating a hard memory deadline from a soft reasoning deadline, grounded in the bounded, passive nature of sensory and working memory. We propose a two-deadline architecture: an always-on, bounded, non-semantic fast path; an active remembering mechanism that commits a moment to durable memory before it is overwritten (event-driven in v1); and a deferred deliberation layer (multimodal model + tools). We instantiate the contract on real MOT17 track windows, a consistency check, and calibrate the single-path baseline from measured frontier-model latency: even a short answer takes 1.2-9.3 s, and an agentic identify-and-search longer. Answers can be late; evidence cannot be missed.

Abstract: Physical systems and distributed AI workflows can retain past information, receive delayed feedback and combine conflicting signals. These features pose a common modeling problem: determining what information about past interactions a state must retain to account for future responses. Active Neural Matter (ANM) addresses this problem through a finite-window response framework. Forward analysis uses each system's dynamics to predict state and output responses to perturbations, separating propagation from output sensitivity. Inverse analysis tests whether a candidate state retains enough information for these responses. With the system and future inputs or interaction rules fixed, distinguishable responses from histories assigned the same state identify distinctions the representation must preserve. Their distributional separation gives a lower bound on the worst-case error of approximating behavior from that state. Uncertainty bounds distinguish rejection, support within a fixed tolerance on tested comparisons, and unresolved cases. Delayed-consensus simulations check analytical predictions of response amplitude and phase over a narrow range of delays; flocking illustrates dependence on the chosen observable. An exact delayed-return construction establishes a sufficient state for a specified program. The AI studies identify distinctions in interaction history and evidence order that a representation must preserve to reproduce observed response distributions. Tests on new histories in a controlled feedback task provide bounded support for a state specified by the task instructions; the instructed rule predicts every observed action. ANM connects response prediction to measurable information requirements, separating behavioral fidelity from task correctness and initial-state adequacy from a closed evolution law for that state.

Abstract: Vision-Language-Action (VLA) models have emerged as powerful foundations for robotic manipulation, but their reliance on fixed camera configurations during training makes them brittle to changes in camera count or pose during deployment. To overcome these limitations, we propose VersaCamVLA, a camera-configurable framework that decouples camera-set representation from action learning. VersaCamVLA learns a unified scene-token interface that maps an arbitrary, variable set of posed RGB views into fixed-size latent scene tokens. This is achieved via multi-signal target-view prediction and Wrist-Augmented Pose Sampling (WAPS), which leverages natural wrist-camera motion for free pose diversity. At deployment, a lightweight spatial encoder injects these compact scene tokens into a pretrained base VLA as a supplementary visual condition, requiring no explicit 3D sensing or novel-view rendering. Experiments on RoboTwin, LIBERO, and a real-robot platform demonstrate that VersaCamVLA consistently outperforms prior VLA methods and direct multi-view baselines, maintaining robust performance across varying camera counts and unseen camera poses.

Abstract: Autonomous rovers navigating large unstructured environments need efficient global planning that accounts for terrain traversability. However, searching dense grid-based costmaps becomes computationally expensive as the mapped area grows. We introduce STAG, a Sparse Traversability-Aware Graph that converts costmaps into compact graphs. STAG combines a medial-axis topological backbone, representative nodes for homogeneous traversability regions, and transition nodes near strong traversability gradients. Edges encode geometry and traversability to account for path length and terrain difficulty. We compare A* on STAG and dense grids using synthetic cave maps, mine maps and the DARPA CERBERUS dataset. Across five benchmark categories comprising 203 map instances and 101,200 queries, STAG reduces median planning time by 3.4x to 9.9x and peak query memory by 2.1x to 15.4x, with median relative path-length differences of -2.9% and +7.6%. STAG offers a compact representation for global planning, trading dense-grid traversability optimality for faster, less memory-intensive search.

Abstract: Goal-conditioned dynamic manipulation of deformable linear objects has mainly specified goals as positions for a rope tip to reach. Many tasks, however, depend on how the tip arrives. We therefore study single-swing rope striking with goals that specify the tip's 3D position and arrival direction, across the workspace and on different ropes. This is challenging because rope dynamics are hard to model, no demonstrations exist, distinct swings reach the same goal with different reliability, and the sim-to-real gap extends beyond the rope. To address these challenges, we extend the state-of-the-art DLO simulator DeformX with GPU acceleration, a stable Cosserat rod solver, and a cross-flow aerodynamic model, yielding DeformX2.0, which is more than $20{,}000\times$ faster. We then propose TRACE (Trace-rooted Adaptive Cross-Entropy), which generates striking data by warm-starting each new target from the stored swing whose tip path passes closest to it. Its cost penalizes rope bending and abrupt tip motion to favor repeatable swings. A conditional flow-matching policy trained on this data reaches 92.1% accuracy in simulation. Finally, we propose RECAP (Residual Calibration Policy), which fits the simulator's rope and rig parameters to a few calibration swings and adapts actions with a correction policy trained in simulation. On a real robot, across three ropes, RECAP raises success within 5cm from 72% to 87% for position goals, and within 10cm and 10° from 50% to 79% for goals that also specify the arrival direction.

Abstract: Adapting vision-language-action (VLA) models to deployment-time distribution shifts is important for reliable robotic operation, but conventional first-order adaptation can exceed the memory budget of inference-oriented deployment platforms. Zeroth-order (ZO) optimization offers a forward-only alternative with inference-level memory, but accurate gradient estimation requires many perturbation queries, making naive ZO prohibitively slow for large VLA models. We present VLA-ZO, a framework for fast ZO adaptation that exploits the structure of VLA computation. By confining adaptation to the action side, VLA-ZO keeps the expensive vision-language prefix frozen and reuses its conditioning states across perturbation queries and optimizer steps, while schedule-aware prefetching hides state-transfer overhead. On LIBERO camera-viewpoint shifts, VLA-ZO reduces end-to-end adaptation time by 25.59$\times$ at $q=16$ and 32.54$\times$ at $q=64$ relative to baseline ZO, while improving average task success from 48.27% without adaptation to 58.17% and 63.58%, respectively. These results show that making ZO faster can make larger query budgets practical, providing a promising path toward resource-efficient VLA adaptation on deployment platforms.

Abstract: Multimodal imitation learning requires diverse executable futures under the same observation and consistent behavior across replanning cycles. We present Conditional Trajectory Peaks (CTP), a single-pass policy framework that jointly predicts complete action-chunk candidates, probability masses, and trajectory scales. Distribution-Aware Peak Specialization (DAPS) specializes trajectory peaks using trajectory-level posterior responsibilities and mass- and scale-modulated overlap constraints. Evidence-Gated Trajectory Belief Transport (ETBT) maintains cross-chunk consistency through geometric correspondence between exchangeable candidate sets, while allowing current policy evidence to override historical constraints. CTP achieves a coverage score of 91.40% on Push-T; success rates of 100.0%, 79.72%, and 84.44% on D3IL Avoiding, Aligning, and Sorting-2, respectively. On LIBERO, CTP achieves an average success rate of 97.25%. In real-world dual-arm experiments, CTP preserves both placement modes in a two-plate task, succeeding in all 50 trials. On bottle uprighting and pen placement into a holder, it maintains success rates comparable to $π_{0.5}$ while reducing policy inference latency from 218.24 ms to 75.80 ms. These results demonstrate that single-pass trajectory modeling can combine multimodal behavior, closed-loop consistency, and efficient inference.

Abstract: Hierarchical reinforcement learning uses temporally extended subtasks for exploration, yet committing to their execution can restrict both deployment and policy learning. We identify and separate the resulting execution and policy suboptimality. Task and execution trees distinguish reward objectives from policy choices and decision interruption. A Unified Value Function for HRL and a four-stage Generalized Hierarchical Bellman Equation then support a common analysis of both losses. Under bounded rewards and uniform termination, we establish hierarchical policy and execution improvement results. With the remaining node policies fixed, task-subtree compatibility and node-policy optimality under the original execution mode establish when Markov execution is optimal. The resulting decomposition leads to independent execution choices for behavior, targets, and deployment. We instantiate this principle through execution improvement and one-stage or two-stage policy improvement at arbitrary hierarchy depth. Option-based and goal-conditioned experiments demonstrate complementary gains from changing execution and changing the learning target. Controlled stochastic environments show how these gains depend on stochastic transition strength and spatial structure. This framework makes execution design an explicit component of hierarchical policy optimization.

Abstract: Language agents should draw on prior attempts and environmental feedback to improve subsequent decisions within the same task. However, providing additional interaction history can sometimes reduce task success, suggesting that agents do not consistently use this information effectively. To investigate this limitation, we examine how agents use history. We find that history improves task completion overall, yet much of this benefit persists even when past actions are shuffled. Disrupting the correspondence between actions and observations causes only a modest decline in task success. We therefore hypothesize that agents do not reliably connect past actions with their outcomes when deciding how to proceed. To test this hypothesis, we explicitly label each returned observation as the outcome of the preceding action. This simple annotation improves task success and reduces next-action repetition without introducing new environmental information. Building on this insight, we introduce a learned calibrator that explicitly reassesses past actions and selectively records experience to guide subsequent decisions, improving task success beyond outcome labeling alone.

Abstract: This paper proposes a communication-free multi-robot task allocation framework based solely on local observations. In this study, tasks are defined as reaching target locations. Each robot estimates the positions of neighboring robots using a Labeled Multi-Bernoulli (LMB) filter and independently assigns tasks through a greedy auction-based strategy. By continuously updating state estimates and reallocating tasks during execution, the proposed method enables decentralized coordination without explicit communication. Monte Carlo simulations demonstrate that the proposed method enables effective cooperative task allocation without inter-robot communication while remaining robust to measurement clutter and observation uncertainty.

Abstract: World models are typically trained and evaluated by prediction error, assuming that more accurate predictions lead to better decisions. We show that this assumption can fail because models with similar total error can differ substantially in planning performance when their errors occur on different state dimensions. We introduce Decision-Relevant Prediction Error (DRPE), which measures prediction error on the state dimensions that affect decisions. We also develop an iso-error evaluation protocol that varies error allocation while keeping total error fixed. In a factored gridworld with known state relevance and a standardized planner, we evaluate 55 controlled and learned models across different error levels and allocations. Total prediction error is weakly related to planning success (Spearman $ρ=-0.25$), whereas DRPE is strongly predictive ($ρ=-0.84$; $-0.98$ within the controlled family). Models with only a 1\% difference in total error can differ by 60 percentage points in planning success (97\% vs 37\%). The relevant error also depends on the task, with model rankings reversing across tasks at the same total error. Deeper imagination further amplifies decision-relevant errors, while learned models exhibit systematic bias on rare but decision-critical events. We formalize sufficient conditions under which DRPE correctly ranks models and total prediction error cannot.

Abstract: We present S2Planner, a trajectory planner that combines three front-facing cameras with ego-motion history and the current driving command. A fine-tuned DINOv3 backbone and a Spatial Tuning Adapter produce multi-scale image features; a coarse-to-fine decoder then uses trajectory self-attention and camera-projected cross-attention to refine candidate waypoints. The contribution is the integration of ego-conditioned trajectory initialization with iterative, geometry-guided sampling of multi-scale image features, rather than a new visual backbone or attention operator. On the NAVSIM v1 non-reactive evaluation, the previously reported navtest run obtained 88.03 PDMS. Because that run was selected using navtest performance, this number is exploratory and cannot be interpreted as an unbiased test estimate. Validation-selected evaluation on unexposed data, repeated runs, and computational measurements are needed to establish generalization and efficiency.

Abstract: Omni-modal models have expanded multimodal interaction across vision, audio, speech, and language. However, their training is predominantly organized around semantic descriptions and general-purpose objectives, leaving physical attributes, interaction states, and causal mechanisms only partially specified. This gap is not simply a matter of modality coverage: adding more modalities does not by itself provide the supervision needed to connect observations with the physical structure of the world. We present OmniFysics-Nano-V2, a compact omni-modal model for physical-world perception and understanding. The model supports image, video, audio, speech, and text inputs within a shared reasoning framework, together with text and speech generation. To address the lack of explicit physical supervision, we construct a dual-branch physics-aware data pipeline that grounds salient objects in structured physical attributes and aligns visual changes with acoustic events, intermediate responses, and interaction outcomes. To address homogeneous training objectives, we curate reinforcement-learning prompts by reward diversity and adopt a two-stage Group Relative Policy Optimization curriculum that progresses from general task correctness to fine-grained physical perceptual reasoning. Experiments across multimodal, audio-visual, and physical reasoning benchmarks show that the proposed data and training strategy improves physical-world understanding while preserving broad omni-modal competence. The proposed model achieves leading result on 17 of 21 benchmarks against SOTA omni-modal models. By equipping AI systems with both omni-modal and physical-world perception capabilities, OmniFysics-Nano-V2 is poised to become a cornerstone of next-generation Physical AI.

Abstract: In constrained motion planning problems, task and loop-closure constraints restrict a robot's motion to a curved, lower-dimensional submanifold of its configuration space. Planners measure path length with a metric, which sets the cost of moving in each direction. Under the Euclidean metric, this cost is the same everywhere, whereas under a general Riemannian metric, such as the kinetic-energy metric, the cost can vary with direction and configuration. Existing methods often describe the submanifold either implicitly, as a constraint level set, or explicitly, through a parameterization. The implicit representation is typically combined with the Euclidean metric of the configuration space, and the explicit representation with the parameter domain, so the path length that a planner minimizes depends on the representation. Instead, we measure path length with the induced metric, which the submanifold inherits from a Riemannian metric on the configuration space. The implicit and explicit representations yield the same induced metric, expressed in different coordinates, and hence the same geometry. This result holds for any Riemannian metric on the configuration space, not only the Euclidean one. The choice of metric is therefore independent of the choice of representation. Using this result, we extend planning under a Riemannian metric from unconstrained spaces to constraint submanifolds by applying the induced metric in both a sampling-based planner and a trajectory optimizer. For an explicit representation, the induced metric also accounts for the distortion that the parameterization introduces. In experiments on a bimanual manipulation setup with two Franka arms under end-effector task constraints, we compare the Euclidean and kinetic-energy metrics.

Abstract: Robots carrying out tasks in dark environments need to localize from a single RGB camera, in light so low that the per-pixel signal approaches the sensor's own noise, on a power-constrained onboard computer, in real time. Each of these constraints has matured pipelines, but the intersection does not. Offline low-light reconstruction now recovers structure below -4 dB but is far too slow to run in real time, while the real-time monocular systems a robot can actually carry (DROID-SLAM, DPV-SLAM, etc.) degrade or fail when SNR gets low. We measured how they fail: across the nine lowest darkness levels of our scenes, DROID-SLAM returns a full-length trajectory carrying no information about the camera's motion on all nine, VGGT-SLAM and CUT3R on eight, pi^3 on seven, and DPV-SLAM on four. We present SYS, a monocular pipeline built on a low-light feed-forward pointmap front end with an explicit match gate, which returns three tracked trajectories and no uninformative ones, at the lowest error of any method where it tracks (24-47% of the no-information ceiling against 56-73% for the strongest baseline), and at the narrowest coverage. On a real robot video take in which 86.5% of delivered frames are entirely black, every configuration of ours stops after the lit beginning, while DROID-SLAM and DPV-SLAM each emit a pose for all 1178 frames. Our method contribution is an embedded execution path for the Jetson AGX Orin: running the map, keyframes and backend at 384 pixels with tracking at 256, together with two fixes to the per-frame pose solve, is a replicated Pareto improvement, 1.28x throughput at 0.964x error on one scene and 1.42x at 0.68x on a second, with 47% less peak GPU memory and 29% less energy per pose. We evaluate on a calibrated, bit-exact regenerable noise ladder, on relabelled real-world dark exposures, and on a new dark-room video ladder recorded from a Boston Dynamics Spot robot.

Abstract: A3P5 NEMESIS is a four-wheel rover intended to combine remote inspection, environmental observation and lightweight manipulation within one serviceable platform. This study develops a photo-constrained geometric reconstruction, a subsystem architecture and a reproducible analytical assessment while distinguishing physical prototype evidence from proposed functions. An exploratory search retrieved 5,000 bibliographic records across ten queries, yielding 4,897 distinct DOI records and 1,212 metadata candidates; selected primary studies and technical documents informed the design. The reconstructed configuration retains the carbon-pattern enclosure, independently steered wheel assemblies, folded manipulator, inclined camera mast and side sampling equipment. A declared 24 kg scenario predicts 3.28 newton-metres of gearbox-output torque per wheel on a 20-degree grade under equal load sharing; a separate static model shows how a 2 kg forward payload reduces the geometric front-tipping bound from 38.1 degrees to 32.7 degrees. These are design screens, not measured operating limits. A public-data calibration benchmark uses 7,344 eligible hourly observations, eight sensor/environmental predictors and chronological training, validation and test partitions. Validation-selected ridge regression achieves a held-out CO root-mean-square error of 0.502 milligrams per cubic metre, with a 95% daily-block bootstrap interval of 0.435-0.569 milligrams per cubic metre. This result concerns an external sensor array and cannot establish NEMESIS accuracy. The combined analysis identifies priority measurements, proposed control interfaces and mission-specific validation requirements. The contribution is a traceable engineering design study and evaluation framework for a prototype whose integrated field performance remains to be established.

Abstract: Humanoid robots often execute motion commands through whole-body controllers (WBCs) that track targets while maintaining balance and stability. However, most WBCs are blind to scene geometry, which can lead to collisions from imperfect target motions that are geometrically unsafe due to perception, planning, or teleoperation errors. We propose RECAL, a Robot--Environment Cross-Attention Layer that wraps a blind WBC to trade off target tracking against collision avoidance using external scene geometry. RECAL supports collision-aware tracking of floating-base and end-effector commands, including collision avoidance for held objects. It represents the robot, held objects, and environment as point clouds, using cross-attention between robot/object points and the environment to produce geometry-aware control features. In simulation, RECAL improves collision avoidance while preserving target-tracking performance across frozen-arm and adaptive-arm locomotion, object-carrying, and standing-manipulation scenarios relative to alternative geometry-aware WBC architectures. We further demonstrate the controller on a real Digit V3 humanoid robot.

Abstract: This paper presents a novel approach to hexapod locomotion by evolving each leg's gait independently through a decentralized evolutionary algorithm. Using the Webots simulator and the Mantis hexapod robot, we optimize individual leg controllers without centralized coordination, allowing emergent behaviors to drive the development of efficient, coordinated locomotion. Our decentralized method is benchmarked against cooperative coevolution, demonstrating improved efficacy in generating stable and adaptive gaits while showing interesting emergent coordination. By enabling independent evolution of leg controllers, this method reduces the complexity of gait optimization and highlights the potential of decentralized strategies for scalable and adaptive robotic systems.

Abstract: Optical coherence tomography is a promising modality for markerless motion tracking due to its high spatial resolution and inherent depth perception. However, existing OCT-based tracking approaches are limited in terms of trackable velocity, particularly when multiple landmarks are tracked sequentially for 6D pose estimation. In this work, we present a predictive tracking approach that propagates positional updates between multiple tracked landmarks to obtain a global pose prediction. This enables more robust tracking under high velocities. Our results demonstrate RMSEs below 1 mm for velocities up to 100 mm/s and up to nine consecutively tracked landmarks, highlighting the potential of global motion propagation and prediction for improving the robustness of OCT-based tracking.

Abstract: Intermediate representations are key to bridging the modality gap between generalizable manipulation policies and large-scale pretrained vision-language models (VLMs). Among these, trajectory-based representations compactly represent motion-relevant cues, yet most existing approaches predict trajectories in 2D image space, resulting in intrinsic 3D ambiguity. Moreover, using 2D trajectories with depth still leaves the free-space waypoints ambiguous, limiting reliable 3D reasoning. To address this, we propose predicting 3D consistent waypoints (3DWay) from multi-view images. By reformulating 3D waypoints prediction as generating multi-view consistent 2D waypoints followed by geometric triangulation, we enable explicit 3D motion specification while preserving the strong priors of pretrained VLMs. The predicted waypoints can guide existing VLA models for better generalization or be directly executed on simple tasks. Extensive experiments show that 3DWay substantially improves 3D spatial grounding and vision-language reasoning, demonstrating strong potential for generalizable robot manipulation. Codes will be released at https://github.com/ziqin-h/3DWay.