Learning from Trials and Errors: Reflective Test-Time Planning for Embodied LLMs
This addresses the challenge of enabling robots to learn from errors during deployment, though it is incremental in enhancing existing embodied LLM frameworks.
The paper tackles the problem of embodied LLMs repeating mistakes without learning from experience by introducing Reflective Test-Time Planning, which integrates reflection-in-action and reflection-on-action to improve decision-making, resulting in significant gains over baseline models on new benchmarks.
Embodied LLMs endow robots with high-level task reasoning, but they cannot reflect on what went wrong or why, turning deployment into a sequence of independent trials where mistakes repeat rather than accumulate into experience. Drawing upon human reflective practitioners, we introduce Reflective Test-Time Planning, which integrates two modes of reflection: \textit{reflection-in-action}, where the agent uses test-time scaling to generate and score multiple candidate actions using internal reflections before execution; and \textit{reflection-on-action}, which uses test-time training to update both its internal reflection model and its action policy based on external reflections after execution. We also include retrospective reflection, allowing the agent to re-evaluate earlier decisions and perform model updates with hindsight for proper long-horizon credit assignment. Experiments on our newly-designed Long-Horizon Household benchmark and MuJoCo Cupboard Fitting benchmark show significant gains over baseline models, with ablative studies validating the complementary roles of reflection-in-action and reflection-on-action. Qualitative analyses, including real-robot trials, highlight behavioral correction through reflection.