A Review of Reward Functions for Reinforcement Learning in the context of Autonomous Driving

A research paper analyzes reward functions used in reinforcement learning for autonomous driving, identifying gaps in current approaches. The study categorizes objectives into Safety, Comfort, Progress, and Traffic Rules compliance, highlighting limitations in objective aggregation and context awareness.

• →Current reward functions for autonomous driving reinforcement learning lack standardization and adequate formulation.
• →The paper categorizes autonomous driving objectives into four main areas: Safety, Comfort, Progress, and Traffic Rules compliance.
• →Existing reward functions struggle with objectives aggregation and are often indifferent to driving context.
• →The research identifies fundamental challenges in developing suitable reward functions for complex autonomous driving scenarios.
• →Future research should focus on reward validation frameworks and context-aware structured rewards that can resolve conflicts.