Effective Trajectory Generation for Robots on General 3D Curved Surface

More and more robots are required to adsorb or crawl on the 3D curved surface in order to assist humans in some dangerous or tedious tasks. Existing methods on curved surface are merely able to plan in 2.5D environment at best, limiting the applications of the robots. In this letter, we propose an effective trajectory generation method on general 3D curved surface. We utilize a pose projection strategy to eliminate terrain contact constraint such that it allows to address the 3D curved surface path planning problem with current optimization solvers. We develop a rational terrain assessment approach based on the local terrain geometry, enabling a direct correlation with the robotic hardware properties. The smoothness and safety of the trajectory on 3D curved surface are significantly improved by optimizing the projected state of the robots. We thoroughly validate our method in various scenarios. The results demonstrate outstanding performance compared to state-of-the-art methods.