DRL-Based Trajectory Tracking for 4WS Robot

Deep Reinforcement Learning-Based Trajectory Tracking Framework for 4WS Robot Considering Switch of Steering Modes

Paper: The 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems & Automation (IROS 2025)

Code Repo: Preparing, Comming soon

Abstract: We proposes a multi-modal trajectory tracking method for 4WS robots considering steering mode switching. The approach decomposes trajectory tracking into mode decision-making and tracking control, both designed based on deep reinforcement learning. The mode decider selects appropriate steering modes based on trajectory information, while the tracker executes motion control. A target trajectory random generator and training environment are developed for data-driven model training.

Geodesic-Based Path Planning for Port Robots

Geodesic-Based Path Planning for Port Transfer Robots on Riemannian Manifolds

Paper: Expert Systems With Applications (ESWA 2025)

Code Repo: https://github.com/balmung08/Geodesic-Based-Path-Planning

Abstract: We propose a geodesic-based path planning method formulated on Riemannian manifolds for port transfer robots. The approach constructs a Riemannian metric tensor that jointly encodes directional motion constraints, steering effort, and obstacle accessibility boundaries into local path cost structures. This transforms the planning task into a geodesic shortest path problem, which is efficiently solved using the Geometric Heat Flow (GHF) method. The resulting paths naturally comply with kinematic constraints and exhibit strong obstacle-avoidance capabilities.