We present a novel algorithm for smooth and collision-free navigation for multiple human-like robots. Our approach combines reciprocal collision avoidance with kinematic and dynamic stability constraints to compute a non-oscillatory trajectory for each high-DOF robot. We use a multi-level optimization algorithm that combines acceleration-velocity obstacles with trajectory optimization. We highlight our algorithm's performance in different environments containing multiple human-like robots with tens of DOFs.


Position Exchange: Two robots exchange their positions on uneven ground.

>Dynamic Obstacles: Robots have to cross moving obstacle's path to navigate to their goals. Circle: 8 robots on a circle move to their goal position opposite their initial position. Narrow Passage: Robots move through a narrow passage which is like a building entrance.


Smooth and Dynamically Stable Navigation of Multiple Human-Like Robots
Chonhyon Park and Dinesh Manocha
The Eleventh International Workshop on the Algorithmic Foundations of Robotics (WAFR), 2014 [PDF]

Related links

Parallel Optimization-based Motion Planning in Dynamic Environments
Reciprocal Collision Avoidance with Acceleration-Velocity Obstacles
Motion Planning Research at GAMMA
GAMMA Research Group