Objective With the development of intelligent coal mining, automated underground bolting is critical to enhance roadway excavation efficiency and safety, while drilling arm path planning faces low automation, tedious posture adjustment, poor stability and inadequate drilling accuracy in complex underground conditions.

Method To address the above issues, an improved rapidly exploring random tree (RRT) algorithm is proposed for intelligent path planning of coal mine rock bolting drilling arm. First, a goal-biased guidance strategy is adopted to narrow the sampling range, an updated list based on collision detection is established to achieve efficient adaptive adjustment of step size, and an adaptive direction weight updating strategy is proposed to realize flexible control of sampling directions. A joint constraint model of the drill boom is constructed to efficiently solve the inverse kinematics of the unconventional multi-DOF manipulator, and constraint criteria are incorporated into the sampling process for real-time restriction and dynamic modification of sampling nodes. Finally, redundant node pruning and a path smoothing strategy are adopted to optimize the path quality. Combined with the solution of the joint constraint model, a globally optimal path planning scheme is obtained, which ensures that the drilling rig always operates perpendicularly to the roadway roof, thus achieving the simultaneous improvement of drilling accuracy and operational efficiency.

Results and Conclusion  Compared with the RRT algorithm and its variants, the proposed algorithm achieves optimal performance in 3D obstacle-avoidance path planning, with a search time of 0.18 s, a final path length of 338.04 mm, and a success rate of 100%. With the introduction of joint constraint, the path length of the drill arm is reduced by 19.71%, and the search time is shortened by 48.89%. In drilling arm planning experiments, its path length is reduced by 45.19% and time consumption is decreased by 15.54% compared with RSA-RRT. Physical experiments verify that the position deviation of the drilling arm end-effector is 1.1~2.9 cm and the drilling angle error is 1.9°~3.9°, which realizes stable and reliable drilling accuracy with superior comprehensive performance and engineering application value.