Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LIVEPOINT: Fully Decentralized, Safe, Deadlock-Free Multi-Robot Control in Cluttered Environments with High-Dimensional Inputs

Jeffrey Chen, Rohan Chandra

TL;DR

The paper addresses decentralized, safe, and deadlock-free multi-robot navigation in cluttered environments using point-cloud data. It introduces LIVEPOINT, which combines universal safety-liveness certificates based on Control Barrier Functions with a symmetry-aware deadlock prevention mechanism that perturbs velocity minimally. The approach achieves zero collisions and 100% success in doorway and intersection scenarios, outperforming ORCA, MPC-CBF, and MPNet in these constrained settings, while maintaining smoother trajectories. The work advances practical multi-robot coordination by enabling real-time safety and liveness guarantees directly from high-dimensional perception, with potential for scaling to more agents and alternative sensing modalities.

Abstract

Fully decentralized, safe, and deadlock-free multi-robot navigation in dynamic, cluttered environments is a critical challenge in robotics. Current methods require exact state measurements in order to enforce safety and liveness e.g. via control barrier functions (CBFs), which is challenging to achieve directly from onboard sensors like lidars and cameras. This work introduces LIVEPOINT, a decentralized control framework that synthesizes universal CBFs over point clouds to enable safe, deadlock-free real-time multi-robot navigation in dynamic, cluttered environments. Further, LIVEPOINT ensures minimally invasive deadlock avoidance behavior by dynamically adjusting agents' speeds based on a novel symmetric interaction metric. We validate our approach in simulation experiments across highly constrained multi-robot scenarios like doorways and intersections. Results demonstrate that LIVEPOINT achieves zero collisions or deadlocks and a 100% success rate in challenging settings compared to optimization-based baselines such as MPC and ORCA and neural methods such as MPNet, which fail in such environments. Despite prioritizing safety and liveness, LIVEPOINT is 35% smoother than baselines in the doorway environment, and maintains agility in constrained environments while still being safe and deadlock-free.

LIVEPOINT: Fully Decentralized, Safe, Deadlock-Free Multi-Robot Control in Cluttered Environments with High-Dimensional Inputs

TL;DR

The paper addresses decentralized, safe, and deadlock-free multi-robot navigation in cluttered environments using point-cloud data. It introduces LIVEPOINT, which combines universal safety-liveness certificates based on Control Barrier Functions with a symmetry-aware deadlock prevention mechanism that perturbs velocity minimally. The approach achieves zero collisions and 100% success in doorway and intersection scenarios, outperforming ORCA, MPC-CBF, and MPNet in these constrained settings, while maintaining smoother trajectories. The work advances practical multi-robot coordination by enabling real-time safety and liveness guarantees directly from high-dimensional perception, with potential for scaling to more agents and alternative sensing modalities.

Abstract

Fully decentralized, safe, and deadlock-free multi-robot navigation in dynamic, cluttered environments is a critical challenge in robotics. Current methods require exact state measurements in order to enforce safety and liveness e.g. via control barrier functions (CBFs), which is challenging to achieve directly from onboard sensors like lidars and cameras. This work introduces LIVEPOINT, a decentralized control framework that synthesizes universal CBFs over point clouds to enable safe, deadlock-free real-time multi-robot navigation in dynamic, cluttered environments. Further, LIVEPOINT ensures minimally invasive deadlock avoidance behavior by dynamically adjusting agents' speeds based on a novel symmetric interaction metric. We validate our approach in simulation experiments across highly constrained multi-robot scenarios like doorways and intersections. Results demonstrate that LIVEPOINT achieves zero collisions or deadlocks and a 100% success rate in challenging settings compared to optimization-based baselines such as MPC and ORCA and neural methods such as MPNet, which fail in such environments. Despite prioritizing safety and liveness, LIVEPOINT is 35% smoother than baselines in the doorway environment, and maintains agility in constrained environments while still being safe and deadlock-free.

Paper Structure

This paper contains 28 sections, 2 theorems, 18 equations, 14 figures, 1 table.

Table of Contents

  1. INTRODUCTION
  2. Main Contributions
  3. Organization of Paper
  4. Related Work
  5. Vision-Based Navigation
  6. Safety and Liveness in Cluttered Environments
  7. PROBLEM FORMULATION
  8. Social Mini Games
  9. BACKGROUND
  10. Control Barrier Functions
  11. Single Robot Navigation with Point Clouds
  12. LivePoint: UNIVERSAL SAFETY AND LIVENESS CERTIFICATES
  13. Safety
  14. Liveness
  15. Deadlock Detection
  16. ...and 13 more sections

Key Result

Theorem 1

Consider a symmetric SMG, such as the one in Figure smg, with two robots length $l$ that are at distance $d$ from the point of collision $\mathcal{Q}$. For the robots to reach $\mathcal{Q}$ without colliding, one of the robots must slow down (or speed up) by a factor of $\zeta \geq 2$ in the limit a

Figures (14)

  • Figure 1: Our decentralized approach enables effective, safe, live, and socially compliant robot navigation in cluttered environments using only high-dimensional point cloud input.
  • Figure 2: Example of a Social Mini-Game. The two robots (red and blue) have straight-line preferred trajectories that intersect at $t=2$, as both must pass through $\mathcal{Q}$ on the way to their goal states, $X^1_g$ and $X^2_g)$. Without predefined rules or corrective actions, the two robots will collide.
  • Figure 3: Technical flowchart illustrating our multi-agent navigation approach. Our universal safety-liveness certificate processes point cloud input to generate robot velocities that ensure deadlock-free navigation. Collectively, the blue elements represent one simulation step.
  • Figure 4: Doorway: Multi-Agent Robot Trajectories with Liveness
  • Figure 5: Doorway: Multi-Agent Robot Trajectories, No Liveness
  • ...and 9 more figures

Theorems & Definitions (7)

  • Definition 1
  • Theorem 1
  • proof
  • Definition 2
  • Definition 3
  • Theorem 2
  • proof