Causal-PIK: Causality-based Physical Reasoning with a Physics-Informed Kernel
This addresses the problem of efficient physical reasoning for agents in interactive environments, representing an incremental improvement through a novel hybrid method.
The paper tackles the problem of planning in tasks with complex object interactions and unknown dynamics by proposing Causal-PIK, a method that uses Bayesian optimization with a Physics-Informed Kernel to guide action search, resulting in outperforming state-of-the-art on Virtual Tools and PHYRE benchmarks with fewer actions to reach goals and remaining competitive with humans on challenging tasks.
Tasks that involve complex interactions between objects with unknown dynamics make planning before execution difficult. These tasks require agents to iteratively improve their actions after actively exploring causes and effects in the environment. For these type of tasks, we propose Causal-PIK, a method that leverages Bayesian optimization to reason about causal interactions via a Physics-Informed Kernel to help guide efficient search for the best next action. Experimental results on Virtual Tools and PHYRE physical reasoning benchmarks show that Causal-PIK outperforms state-of-the-art results, requiring fewer actions to reach the goal. We also compare Causal-PIK to human studies, including results from a new user study we conducted on the PHYRE benchmark. We find that Causal-PIK remains competitive on tasks that are very challenging, even for human problem-solvers.