G-MAPP: GPU-accelerated Multi-Agent Planning and Perception for Reactive Motion Generation

Tanmay Bishnoi1, Riddhiman Laha2,3,4, Tobias Löw5,6, Jose Alex Chandy7,8, Luis F. C. Figueredo3,7,8, Sami Haddadin9
1Toronto Metropolitan University  ·  2Northeastern University  ·  3MIRMI, TU Munich  ·  4Technical University of Munich  ·  5Idiap Research Institute  ·  6EPFL  ·  7CHART Research  ·  8University of Nottingham  ·  9MBZUAI
Paper arXiv Video Code
G-MAPP framework overview

g-mapp is a coupled real-time GPU-accelerated perception and planning framework. It utilizes a fast perception pipeline, a multi-agent planner, and a vector-field engine to generate robot trajectories.

Abstract

Reactive motion generation in unstructured environments remains an open challenge in robotics. Due to the computational complexity of collision-free motion generation, existing methods either generate global trajectories for static scenarios, or employ models that make conservative assumptions about the environment.

This paper identifies the primary bottleneck as the runtime performance demand of planning on high-fidelity environments, and the temporal integration between the perception and planning modules. Therefore, we propose a framework that does not compromise on runtime performance and world representations for perception and planning by accelerating world modeling and vector-field based planning using the GPU. This allows us to achieve faster parallel state exploration for quasi-global trajectory planning, and tighter coupling of the perception-action loop in real-time for dynamic cluttered environments with off-the-shelf depth sensors.

We quantitatively evaluate the computation-time and success rate differences for the CPU and GPU versions of our planner, and perform qualitative evaluations of our coupled framework using real-world experiments on a 7-DoF Franka Emika robot. Experimental results demonstrate that our GPU-based framework achieves up to a 5x speedup over the CPU in non-trivial scenes.

Simulations

Real-World Experiments

Static Concave Trap

This scenario features a static environment, evaluating performance in predictable settings with local minima.

Dynamic Obstacle Avoidance

This scenario tests the ability to generate collision-free paths in response to unexpected changes in environment.

Video

Installation

Our framework is available on the CHART-Research/g-mapp github under the MIT license for free use, research, and development by the robotics community.

BibTeX

@article{bishnoi2026gmapp,
  author={Bishnoi, Tanmay and Laha, Riddhiman and Löw, Tobias and Chandy, Jose Alex and Figueredo, Luis F. C. and Haddadin, Sami},
  journal={IEEE Robotics and Automation Letters}, 
  title={G-MAPP: GPU-Accelerated Multi-Agent Planning and Perception for Reactive Motion Generation}, 
  year={2026},
  volume={11},
  number={6},
  pages={7516-7523},
  doi={10.1109/LRA.2026.3678839}
}