Star-World reshaping boosts Franka arm motion planning success rate by 17%

Safety inflation in reactive motion planning often causes nearby obstacles to overlap, violating the disjoint-obstacle assumptions that many modulation-based planners rely on. To address this, the team from (unclear affiliation, likely academic) proposes Star-World workspace reshaping for 3D control of a Franka Emika Panda manipulator. At each update cycle, intersecting inflated obstacles are clustered and replaced by star-shaped convex proxies. A dynamical-system-based end-effector controller then computes motions, while a null-space artificial-potential-field term provides complementary arm-body avoidance. This approach is evaluated against an unreshaped baseline in six PyBullet scenarios measuring goal attainment, path-length ratio, and computation time.

In this preliminary 12-trial evaluation, the reshaped planner reached the goal in five of six scenarios compared to four for the baseline, successfully resolving the canonical overlapping-wall case. Per-workspace update times ranged from 0.68 ms for a single obstacle to 8.70 ms for seven obstacles. However, the method introduced trade-offs: path lengths increased, two cases exhibited near-equilibrium behavior, and over-aggressive merging closed a navigable corridor. These results demonstrate both the potential of transferring Star-World guarantees to redundant manipulators controlled through inverse kinematics and the practical limitations that must be addressed for real-world deployment.