Action-conditioned Benchmarking of Robotic Video Prediction Models: a Comparative Study
This work addresses the need for better benchmarking of video prediction models in robotics, though it is incremental as it introduces a new evaluation method rather than a novel prediction model.
The paper tackles the problem of evaluating video prediction models for robotic planning by proposing a new metric based on action inference from predicted frames, showing that models with high perceptual scores can perform poorly in action inference tests.
A defining characteristic of intelligent systems is the ability to make action decisions based on the anticipated outcomes. Video prediction systems have been demonstrated as a solution for predicting how the future will unfold visually, and thus, many models have been proposed that are capable of predicting future frames based on a history of observed frames~(and sometimes robot actions). However, a comprehensive method for determining the fitness of different video prediction models at guiding the selection of actions is yet to be developed. Current metrics assess video prediction models based on human perception of frame quality. In contrast, we argue that if these systems are to be used to guide action, necessarily, the actions the robot performs should be encoded in the predicted frames. In this paper, we are proposing a new metric to compare different video prediction models based on this argument. More specifically, we propose an action inference system and quantitatively rank different models based on how well we can infer the robot actions from the predicted frames. Our extensive experiments show that models with high perceptual scores can perform poorly in the proposed action inference tests and thus, may not be suitable options to be used in robot planning systems.