R. Gomm, V. Bhaskar, and S. Cetinkunt
Mobile construction machinery, sampling-based motion planning, heuristic-driven search, adaptive trajectory generation
We present a motion planning algorithm for a linkage mechanism with holonomic constraints as found in mobile construction machinery. It computes in real-time collision-free paths in configuration space (C-Space) for the five degree of freedom (DOF) linkage mechanism into desired positions in work space (W-Space). Our approach is general and based on an approximation of the collision-free portion of the C-Space obtained using two-level resolution sampling. There are two major requirements for the algorithm: (1) It must converge in real-time reliably. (2) It must be implementable with current and future anticipated embedded computer resources on construction machinery. To satisfy these two major demands, our motion planner is realized as a two-phase approach comprising of off-line construction of five-dimensional static C-Space map and on-line heuristic-driven path search, trajectory generation and adaptive control. Machine steering and articulation are decoupled to ensure full machine drivability at all times and to reduce problem complexity to lower memory requirement and computation time. Simulation and experimental results obtained from actual construction equipment machinery show that path and trajectory planning can be accomplished in real time on the target application.
Important Links:
Go Back
