# Coaching a Teachable Student

Source: [Coaching a Teachable Student](<https://www.sota2.com/research/paper/68fa6fca1e209131cea4768a>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-04T03:52:51.042Z

Authors: Jimuyang Zhang, Zanming Huang, Eshed Ohn-Bar

Published: 2023-06-16T17:59:38.000Z

## Abstract

We propose a novel knowledge distillation framework for effectively teaching a sensorimotor student agent to drive from the supervision of a privileged teacher agent. Current distillation for sensorimotor agents methods tend to result in suboptimal learned driving behavior by the student, which we hypothesize is due to inherent differences between the input, modeling capacity, and optimization processes of the two agents. We develop a novel distillation scheme that can address these limitations and close the gap between the sensorimotor agent and its privileged teacher. Our key insight is to design a student which learns to align their input features with the teacher's privileged Bird's Eye View (BEV) space. The student then can benefit from direct supervision by the teacher over the internal representation learning. To scaffold the difficult sensorimotor learning task, the student model is optimized via a student-paced coaching mechanism with various auxiliary supervision. We further propose a high-capacity imitation learned privileged agent that surpasses prior privileged agents in CARLA and ensures the student learns safe driving behavior. Our proposed sensorimotor agent results in a robust image-based behavior cloning agent in CARLA, improving over current models by over 20.6% in driving score without requiring LiDAR, historical observations, ensemble of models, on-policy data aggregation or reinforcement learning.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Autonomous Driving on CARLA Town02 & Town05 (Metrics: DS, RC, IS)](<https://www.sota2.com/research/sota/autonomous-driving-on-carla-town02-and-town05-metrics-ds-rc-is.md>): Autonomous Driving; CARLA Town02 & Town05; DS: 66.7.

- [Autonomous Driving on Longest6 36 routes 1.0](<https://www.sota2.com/research/sota/autonomous-driving-on-longest6-36-routes-1-0.md>): Autonomous Driving; Longest6 36 routes 1.0; DS: 0.5836.

- [Open-loop planning on nuScenes (val)](<https://www.sota2.com/research/sota/open-loop-planning-on-nuscenes-val.md>): Open-loop planning; nuScenes (val); L2 Error (3s): not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2306.10014>)

[PDF](<https://arxiv.org/pdf/2306.10014.pdf>)