Conversational Code Generation: a Case Study of Designing a Dialogue System for Generating Driving Scenarios for Testing Autonomous Vehicles
This work addresses the challenge for non-coding domain experts in creating simulation scenarios for autonomous vehicle testing, though it is incremental as it applies existing methods to a specific domain.
The paper tackled the problem of generating driving scenarios for testing autonomous vehicles by designing a dialogue system that uses a large language model to convert natural language instructions into symbolic programs, achieving a 4.5 times higher success rate with dialogue compared to non-conversational generation.
Cyber-physical systems like autonomous vehicles are tested in simulation before deployment, using domain-specific programs for scenario specification. To aid the testing of autonomous vehicles in simulation, we design a natural language interface, using an instruction-following large language model, to assist a non-coding domain expert in synthesising the desired scenarios and vehicle behaviours. We show that using it to convert utterances to the symbolic program is feasible, despite the very small training dataset. Human experiments show that dialogue is critical to successful simulation generation, leading to a 4.5 times higher success rate than a generation without engaging in extended conversation.