RMMI: Enhanced Obstacle Avoidance for Reactive Mobile Manipulation using an Implicit Neural Map
🤖 AI Summary
Mobile manipulation robots face base-arm coordinated collision avoidance challenges in cluttered static environments. Existing reactive control methods rely on simplified geometric representations, compromising safety and robustness in complex scenarios. This paper proposes the first real-time reactive control framework based on implicit neural signed distance fields (SDFs): it directly embeds neural SDFs into quadratic programming (QP) optimization, constructs differentiable inequality constraints using environmental SDF gradients, and introduces a novel proactive collision-avoidance cost term that maximizes the minimum obstacle distance. The method integrates sensor-driven online reconstruction, joint kinematic modeling, and end-to-end navigation-manipulation optimization. In simulation, task success rate improves by 25%—with proactive avoidance contributing 10%—while real-world experiments demonstrate stable achievement of target poses in narrow, cluttered spaces. The model operates directly on raw sensor data.
Application Category
📝 Abstract
We introduce RMMI, a novel reactive control framework for mobile manipulators operating in complex, static environments. Our approach leverages a neural Signed Distance Field (SDF) to model intricate environment details and incorporates this representation as inequality constraints within a Quadratic Program (QP) to coordinate robot joint and base motion. A key contribution is the introduction of an active collision avoidance cost term that maximises the total robot distance to obstacles during the motion. We first evaluate our approach in a simulated reaching task, outperforming previous methods that rely on representing both the robot and the scene as a set of primitive geometries. Compared with the baseline, we improved the task success rate by 25% in total, which includes increases of 10% by using the active collision cost. We also demonstrate our approach on a real-world platform, showing its effectiveness in reaching target poses in cluttered and confined spaces using environment models built directly from sensor data. For additional details and experiment videos, visit https://rmmi.github.io/.
Problem
Research questions and friction points this paper is trying to address.
Reactive mobile manipulation in cluttered environments
Avoiding collisions using neural implicit maps
Coordinating base and arm motion while maximizing manipulability
Innovation
Methods, ideas, or system contributions that make the work stand out.
Reactive control framework using neural SDFs
Quadratic program optimizing base and arm motion
Collision avoidance via SDF query constraints
🔎 Similar Papers
No similar papers found.
