This work introduces Bamboo GPU Mode, a game-theoretic evolutionary framework designed for resilient autonomous navigation in extreme planetary environments. The proposed method enables autonomous systems to reconstruct coordinated navigation behavior after partial or complete system failure. By integrating sigmoid-parameterized payoff stabilization, phototropic evolutionary mutation, and GPU-parallel computation, the framework demonstrates robust self-healing performance in Mars-like Digital Elevation Model environments under severe sensor noise and terrain disturbance. Simulation results indicate stable recovery behavior and high coordination convergence under conditions where conventional navigation and optimization-based methods tend to fail.
hiroaki ueshima (Sat,) studied this question.