SwapTransformer: highway overtaking tactical planner model via imitation learning on OSHA dataset

TL;DR

This work tackles high-level highway overtaking and lane-change decisions for Travel Assist using imitation learning. It introduces SwapTransformer, a dimension-swapping transformer that fuses lane-ID segmentation images with object-list features over a 5-second horizon, augmented by auxiliary tasks for future pose and car-distance reasoning. Trained on the OSHA dataset (~9 million samples from SimPilot), SwapTransformer outperforms MLP and transformer baselines across low, medium, and high traffic densities in simulation, showing faster lap completion, smaller speed deviations, and more overtakes. The OSHA dataset is released publicly to promote reproducibility and advance safer, more robust tactical planning for autonomous driving systems.

Abstract

This paper investigates the high-level decision-making problem in highway scenarios regarding lane changing and over-taking other slower vehicles. In particular, this paper aims to improve the Travel Assist feature for automatic overtaking and lane changes on highways. About 9 million samples including lane images and other dynamic objects are collected in simulation. This data; Overtaking on Simulated HighwAys (OSHA) dataset is released to tackle this challenge. To solve this problem, an architecture called SwapTransformer is designed and implemented as an imitation learning approach on the OSHA dataset. Moreover, auxiliary tasks such as future points and car distance network predictions are proposed to aid the model in better understanding the surrounding environment. The performance of the proposed solution is compared with a multi-layer perceptron (MLP) and multi-head self-attention networks as baselines in a simulation environment. We also demonstrate the performance of the model with and without auxiliary tasks. All models are evaluated based on different metrics such as time to finish each lap, number of overtakes, and speed difference with speed limit. The evaluation shows that the SwapTransformer model outperforms other models in different traffic densities in the inference phase.

Figures (12)

• Figure 1: An example of an ego vehicle overtaking two agents by making left and right lane changes.
• Figure 2: SwapTransformer architecture to interact with Travel Assist controller.
• Figure 3: Time and feature swapping in SwapTransformer.
• Figure 4: Frames of decision-making with the SwapTransformer at the inference time.
• Figure 5: Travel Assist flowchart including different states.