New framework guides robots through complex terrain using LLM-MPC

Researchers have developed ASTRIL-MPC, a novel framework for autonomous traversal in articulated tracked robots (ATRs) designed for urban search and rescue missions. This system integrates a language-guided neural kinematics model predictive control (MPC) approach, enabling robots to navigate complex environments like stairwells and cluttered interiors. The framework utilizes a learned kinematics model for short-horizon predictions, NMPC for planning with constraints, and a large language model (LLM) for bounded updates to control parameters, achieving a full control cycle within 100 ms. ASTRIL-MPC demonstrated significant improvements in traversal quality, outperforming non-adaptive NMPC and PPO baselines, while also eliminating collision impacts during descent. AI

IMPACT This framework could enhance the autonomy and safety of robots in challenging environments like disaster zones.

RANK_REASON This is a research paper describing a new framework for robots. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Articulated Tracked Robots
• ASTRIL-MPC
• large language model
• Proximal Policy Optimization
• Zhenfeng Gan

• paper
• product
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →