Plan Right, Then Plan Tight: Symbolic RL for Efficient Embodied Reasoning

2026-06-30

AI Summary

This work addresses the scarcity of efficient, verifiable supervisory signals in embodied task planning by proposing a unified framework grounded in Behavior Description Domain Language (BDDL) specifications. It introduces, for the first time, automatically generated BDDL for data construction, plan verification, and reward design. The approach integrates a video-to-BDDL parser, a large language model–based verifier, and a lightweight symbolic reasoning engine to deliver millisecond-level dense feedback. Additionally, it incorporates GroupAdapt, a zero-overhead, difficulty-aware sequence-length scheduling strategy that substantially enhances planning efficiency. Evaluated on BEHAVIOR-1000, the method achieves a Strict-Pass rate of 97.3%, outperforming the Qwen3-8B baseline by 25.9% while reducing response length by 79% to just 207 tokens.

Abstract

Embodied task planning asks an agent to turn a natural-language instruction into an executable sequence of actions in a physical scene, and is a building block for household, assistive, and service robots. Recent prompting-based and reinforcement-learning planners generate fluent action text but lack a cheap deterministic check that the produced plan is valid in the target world, while high-fidelity simulation is too slow to serve as an inner-loop training signal. The general problem is therefore how to obtain verifiable supervision and rewards for embodied planners without relying on string-level matching or full simulation. Here we show that a single BDDL specification, automatically constructed from open-world video evidence or curated tasks, can serve as a shared interface for data construction, plan verification, and reward design. A video-to-BDDL parser, an LLM verifier, and a lightweight symbolic engine together supply dense feedback at millisecond latency. We further introduce GroupAdapt, a difficulty-aware length schedule that uses the in-batch group pass rate as a zero-cost signal so that hard prompts get wider length tolerance and automatically tighten as their pass rate improves. Under the guidance of the proposed verifier and GroupAdapt schedule, the 8B planner attains a Strict-Pass score of 97.3 on BEHAVIOR-1000, yielding a 25.9 percent relative improvement over the Qwen3-8B baseline. This result exceeds the strongest large-model baseline by 3.5 percent, while simultaneously compressing the response length by 79 percent to 207 tokens, demonstrating both effectiveness and efficiency.

Problem

• embodied planning
• plan verification
• reinforcement learning
• symbolic reasoning
• efficient simulation

Innovation

• Symbolic Reinforcement Learning
• BDDL Specification
• Plan Verification
• GroupAdapt
• Embodied Task Planning