Interpretable Trade-offs Between Robot Task Accuracy and Compute Efficiency

TL;DR

This paper introduces an optimal method for robot model selection that balances task accuracy and compute costs by analyzing the trade-offs between fast, less accurate and slow, more accurate computation models, applicable across various robotic tasks.

Contribution

It presents a novel cost-benefit analysis framework for model selection in robotics, leveraging statistical correlations to optimize accuracy and efficiency trade-offs.

Findings

The approach effectively balances accuracy and compute costs in perception tasks.

It improves task success rates within energy and compute budgets.

Applicable to diverse robotic applications like perception and navigation.

Abstract

A robot can invoke heterogeneous computation resources such as CPUs, cloud GPU servers, or even human computation for achieving a high-level goal. The problem of invoking an appropriate computation model so that it will successfully complete a task while keeping its compute and energy costs within a budget is called a model selection problem. In this paper, we present an optimal solution to the model selection problem with two compute models, the first being fast but less accurate, and the second being slow but more accurate. The main insight behind our solution is that a robot should invoke the slower compute model only when the benefits from the gain in accuracy outweigh the computational costs. We show that such cost-benefit analysis can be performed by leveraging the statistical correlation between the accuracy of fast and slow compute models. We demonstrate the broad applicability of our approach to diverse problems such as perception using neural networks and safe navigation of a simulated Mars rover.