Handling many cooperative robots to move objects with less wear
Distributed Dexterous Manipulation with Spatially Conditioned Multi-Agent Transformers
RoboticsMachine Learning
Summary
Controlling many tiny robots working together to move objects is tricky because they have many ways to act and must coordinate well. The authors used a special type of AI called multi-agent transformers that take into account where each robot is to help them work smoothly together. They trained their system on a grid of 64 soft robots to control objects accurately while using fewer robots and reducing collisions and wear. Their method works both in computer simulations and with real robots, keeping errors low even when using about one-third fewer robots than usual.
Distributed Dexterous Manipulationmulti-agent transformerssoft robotsspatial conditioningadaptive layer normalizationbehavior cloningsoft actor criticrobot cooperationaction selectionlong-horizon manipulation
Authors
Sarvesh Patil
Abstract
Distributed Dexterous Manipulation (DDM) is a novel paradigm that presents significant control challenges due to high action-space redundancy, inter-robot cooperation, and dynamic object-robot interactions. This paper introduces a framework based on spatially conditioned Multi-Agent Transformers (MATs) to efficiently learn robust control policies for a DDM system grounded in an array of 64 soft delta robots arranged in an 8x8 grid. Our three core contributions are: (i) an MAT with adaptive layer norm for compute efficiency, (ii) spatial contrastive embeddings to ground transformer embeddings in the spatial configuration of the robots, and (iii) an MAT-based behavior cloning method fine-tuned using Soft Actor Critic. We also propose an action selection formulation to analyze the trade-off between task performance and the number of robots utilized. Our experiments show that MATs iteratively refine their actions through the stacked attention blocks. This further informs the benefit of spatial conditioning in transformers to learn DDM policies. We demonstrate long-horizon planar manipulation tasks with objects of various geometries in simulation and real-world. Finally, we show how action selection mitigates robot maintenance by reducing wear and tear due to inter-robot collisions while maintaining the ability to manipulate objects along various trajectories in the real-world, achieving an average error of ~1.5 cm, while using ~65% fewer robots.