The invention relates to the technical field of robots, and discloses a complex
terrain inspection
robot path planning and
gait optimization method, which realizes cross-scale parallel
perception of a macroscopic grid and a microscopic
landing point for the first time through multi-scale
terrain perception and
uncertainty quantification, fundamentally solves the fundamental contradiction of macroscopic passability and microscopic inaccessibility, and improves the accuracy of the
route planning and
gait optimization of the complex
terrain inspection
robot. By constructing a multi-constraint collaborative planning and foot-ground
interaction time-varying model, the microcosmic
landing point accessibility and the terrain evolution law are fused into path-
gait joint optimization, and the full-cycle conflict between walking dynamic stability and operation static stability is radically solved. Aiming at the problem of
coupling resonance of contact operation of the foot-type
robot, a cross-domain
coupling dynamic model is established,
millisecond-level real-time solution of an edge end is realized by adopting
model order reduction and a lightweight network,
resonance is actively inhibited, the operation precision is ensured to reach + / -1N / 0.5 mm, a full-link uncertainty nonlinear conduction and risk grading closed-loop correction mechanism is introduced, and the operation precision is ensured to reach + / -1N / 0.5 mm. And self-learning and self-healing of the
system are realized.