A Survey on Approximate Edge AI for Energy Efficient Autonomous Driving Services
Dewant Katare, Diego Perino, Jari Nurmi, Martijn Warnier, Marijn Janssen, Aaron Yi Ding
TL;DR
The paper tackles the challenge of energy-efficient autonomous driving by surveying energy-conscious, approximate Edge AI techniques and datasets. It develops a comprehensive taxonomy spanning AI models, edge computing, federated learning, communications, and energy-efficient frameworks, and surveys both driving frameworks (Autoware, Apollo, etc.) and energy-focused edge platforms (OpenVDAP, CAVBench, π-Edge, LoPECS, AC4AV). Key contributions include a structured overview of perception, HD maps, SLAM, and V2X in the context of edge intelligence, plus detailed coverage of Edge Training, Edge Inference, and Federated Learning for connected vehicles. The practical impact lies in guiding the development of low-power, memory-constrained autonomous driving systems and informing the design of cooperative driving services with energy-aware edge pipelines. Overall, the survey highlights open problems in data management, collaborative edge intelligence, and energy-efficient evaluation, and outlines an envisioned Edge AI pipeline to enable scalable, energy-conscious Level 4+ autonomous driving.
Abstract
Autonomous driving services rely heavily on sensors such as cameras, LiDAR, radar, and communication modules. A common practice of processing the sensed data is using a high-performance computing unit placed inside the vehicle, which deploys AI models and algorithms to act as the brain or administrator of the vehicle. The vehicular data generated from average hours of driving can be up to 20 Terabytes depending on the data rate and specification of the sensors. Given the scale and fast growth of services for autonomous driving, it is essential to improve the overall energy and environmental efficiency, especially in the trend towards vehicular electrification (e.g., battery-powered). Although the areas have seen significant advancements in sensor technologies, wireless communications, computing and AI/ML algorithms, the challenge still exists in how to apply and integrate those technology innovations to achieve energy efficiency. This survey reviews and compares the connected vehicular applications, vehicular communications, approximation and Edge AI techniques. The focus is on energy efficiency by covering newly proposed approximation and enabling frameworks. To the best of our knowledge, this survey is the first to review the latest approximate Edge AI frameworks and publicly available datasets in energy-efficient autonomous driving. The insights and vision from this survey can be beneficial for the collaborative driving service development on low-power and memory-constrained systems and also for the energy optimization of autonomous vehicles.