Mobile robots improve precise multi-object placement with diffusion planning
Non-Prehensile Multi-Object Placement with Mobile Robots via Diffusion-Based Trajectory Planning
Robotics
Summary
Precision placement of multiple objects by mobile robots is challenging because the robots must carefully push objects into tight spots without grabbing them. The authors developed a new planning method that models pushing actions as a sequence where the robot's movements and object responses are considered step-by-step. This approach lets the robot predict multiple possible object motions for each plan and pick the best one. Their experiments show this method places more objects correctly and is faster than alternatives, even working well with objects the robot hasn't seen before.
What this means in practice
- •For warehouse automation teams: Increase the accuracy and speed of placing multiple boxes in tight warehouse spaces using mobile robot pushers.
- •For robotics system integrators: Deploy improved pushing strategies on low-cost mobile robots to handle unfamiliar object shapes precisely without grippers.
- •For automated material handling vendors: Develop enhanced robotic placement solutions for logistics that achieve high-precision multi-object arrangements with mobile robot platforms.$Commercial implications: Enables commercial production of precise multi-object placement robots for logistics companies needing better automation.
Authors
Lance Pharand, Megnath Ramesh, Mark Chiu, Yash Vardhan Pant, Stephen L. Smith
Abstract
Non-prehensile manipulation with mobile robots has primarily targeted placement in coarse goal regions. Precision placement has been studied with manipulators on benchmarks such as Push-T, where an omnidirectional end effector can readily re-approach an object from any side. For a mobile robot, correcting the same placement errors may require wide repositioning maneuvers around the object. We formalize multi-object pushing with a mobile robot as a multi-target assignment problem at tight tolerances. We propose a diffusion planner with a causal factorization: each robot denoising step is followed by an object step conditioned on the current clean robot estimate. This factorization allows us to sample multiple object outcomes for a single robot plan, capturing the multi-modality of pushing. We also design a set of pushing-specific costs for mobile robots, used for both guidance and best-of-$N$ sampling at inference. On Push-T, our planner attains the highest success rate of the methods we compare, while having the lowest rate of implausible object motion. Across three multi-box pushing environments resembling warehouse scenarios, it raises the fraction of boxes placed by $21.5$ percentage points over an unguided baseline, and replans $1.8\times$ faster than a planner that denoises the two trajectories together. Lastly, we deploy the guided planner on a TurtleBot 3 Burger with object shapes absent from the training data, where it places $85.0\%$ of objects within tolerance against Diffusion Policy's $37.5\%$.