Transformer-Based Autonomous Driving Models and Deployment-Oriented Compression: A Survey
This survey paper examines the use of Transformer-based models in autonomous driving systems. It categorizes these models by their task role, sensing configuration, and architecture, while also analyzing how efficiency constraints influence their design. The paper further reviews various compression and acceleration strategies, such as quantization and pruning, that are crucial for deploying these models in real vehicles. Ultimately, it highlights the need to consider compression as a fundamental system-level design element for safety and deployability. AI
IMPACT Provides a structured overview of current research and challenges in applying advanced AI models to autonomous vehicles.