Abstract: During the autonomous navigation process of underground inspection robots in complex and multi-obstacle tunnels,
they often encounter problems such as low path planning efficiency,poor obstacle avoidance stability,and insufficient realtime
performance. To achieve efficient autonomous navigation in complex mine environments,a path planning model integrating
the improved multi-objective salp swarm algorithm (Improved Multi-objective Salp Swarm Algorithm,IMSSA) and the fuzzy
dynamic window approach (Fuzzy Dynamic Window Approach,FDWA) is proposed (IMSSA-FDWA model). By improving the
multi-objective salp swarm algorithm,a chaotic initialization strategy is introduced to enhance population diversity,and an energy
consumption adaptive weighting mechanism is adopted to balance path length and energy loss,thereby improving global
search and path optimization capabilities. On this basis,the fuzzy dynamic window approach is combined to adjust the linear
and angular velocities to achieve real-time obstacle avoidance control in dynamic obstacle environments. Experimental results
show that the average planning time of this model is 0. 67 s,the path length is 57. 8 m,the variance of continuous turning segments
is 0. 016 5 rad2,the path deviation error is 0. 09 m,the obstacle avoidance success rate is 96. 59%,and the average response
delay is 35 ms. Compared with algorithms such as the Improved Particle Swarm Optimization (IPSO),the A∗ with Dynamic
Window Approach (A∗-DWA),and the Artificial Potential Field with Dynamic Window Approach (APF-DWA),the
IMSSA-FDWA model plans smoother paths,has stronger obstacle avoidance stability,and better real-time response capability.
The IMSSA-FDWA model can achieve high-precision path generation and autonomous obstacle avoidance in complex underground
environments,providing a feasible technical path for the safe and efficient operation of intelligent inspection robots in
dynamic and highly constrained roadways.

Key words: mine inspection robot,path planning,multi-objective tunicate algorithm,fuzzy dynamic window method