Failure guided training improves quadruped robot parkour success rates
FCTS: Failure-Guided Curriculum Tree Search Over Generated Terrain for Legged Parkour
Robotics
Summary
Robots that run and jump over obstacles often struggle with complex, irregular environments like construction sites. The authors created a system called FCTS that learns where the robot fails and then trains it on similar tricky terrain automatically. This approach helps the robot get much better at navigating hard courses, even those based on real-world locations. Tests show that robots trained with FCTS do better both in simulation and in the real world.
What this means in practice
- •For robotics engineers: Train legged robots to navigate complex industrial or disaster environments by automatically focusing training on failure points identified in previous attempts.
- •For robot operators: Deploy quadruped robots with improved parkour skills to traverse challenging real-world sites safely and reliably.
Authors
Narayanan Palghat Parameswaran, Dong-Ki Kim, Jay Patrikar, Santiago Gutierrez, Harshad Zade, Shayegan Omidshafiei, Harmish Khambhaita
Abstract
Quadruped parkour policies handle canonical obstacles well, but their training curricula often miss the irregular, compound geometry of industrial and disaster-response sites, where agile locomotion matters most. Such environments pose a hard generalization problem: even training directly on reconstructed geometry from the target site does not reliably yield policies that traverse it. Hand-designed curricula span only a few terrain families, and recent LLM-generated curricula offer little control over which policy weaknesses they target. We propose FCTS, a failure-conditioned terrain curriculum framework that uses localized traversal failures to automatically generate the terrains a policy should train on next. Each failure's location, termination mode, robot state, and surrounding terrain geometry condition this generation, and Monte Carlo Tree Search explores alternative curriculum trajectories under a fixed training budget. On terrains built from scanned real-world construction sites, FCTS raises success rate from 28.4% to 85.8% on a medium-difficulty course and from 14.9% to 44.7% on a hard course relative to the strongest parkour or LLM-curriculum baselines, and maintains higher progress on extreme terrains where full traversal is rare. Hardware experiments on a quadruped robot confirm that the learned FCTS policies transfer to the real world.