Enhanced B-RRT* With Discretized Collision Detection Model for Manipulator in Joint Space

Manipulators are revolutionizing intelligent manufacturing by autonomously performing high-risk operations with precision and reliability. Path planning technology is central to manipulator control, as high-quality paths must be generated to ensure operational safety and efficiency. This study introduces an enhanced bidirectional, rapidly exploring random tree* algorithm that integrates joint-space sampling to eliminate angular discontinuities and a joint-weighted distance to account for energy consumption. The proposed method incorporates a variable-interval discretization strategy, which transforms the complex problem of collision detection into a real-time analysis of positional relationships between discretized points and obstacles. Furthermore, a hybrid sampling and expansion strategy is adopted to improve overall path planning performance. Experiments confirm the algorithm reliably avoids obstacles in complex environments while consistently maintaining safe margins. Moreover, it consumes less energy and requires fewer valid nodes than conventional algorithms.