Recent years have witnessed remarkable progress in autonomous driving, with systems evolving from modular pipelines to end-to-end architectures. However, most existing methods are trained offline and lack mechanisms to adapt to new environments during deployment. As a result, their generalization ability diminishes when faced with unseen variations in real-world driving scenarios. In this paper, we break away from the conventional "train once, deploy forever" paradigm and propose EvoPSF, a novel online Evolution framework for autonomous driving based on Planning-State Feedback. We argue that planning failures are primarily caused by inaccurate object-level motion predictions, and such failures are often reflected in the form of increased planner uncertainty. To address this, we treat planner uncertainty as a trigger for online evolution, using it as a diagnostic signal to initiate targeted model updates. Rather than performing blind updates, we leverage the planner's agent-agent attention to identify the specific objects that the ego vehicle attends to most, which are primarily responsible for the planning failures. For these critical objects, we compute a targeted self-supervised loss by comparing their predicted waypoints from the prediction module with their actual future positions, selected from the perception module's outputs with high confidence scores. This loss is then backpropagated to adapt the model online. As a result, our method improves the model's robustness to environmental changes, leads to more precise motion predictions, and therefore enables more accurate and stable planning behaviors. Experiments on both cross-region and corrupted variants of the nuScenes dataset demonstrate that EvoPSF consistently improves planning performance under challenging conditions.

Portfolio Online Evolution is a novel method for playing real-time strategy games through evolutionary search in the space of assignments of scripts to individual game units. This method builds on and recombines two recently devised methods for playing multi-action games: (1) Portfolio Greedy Search, which searches in the space of heuristics assigned to units rather than in the space of actions, and (2) Online Evolution, which uses evolution rather than tree search to effectively play games where multiple actions per turn lead to enormous branching factors. The combination of both ideas lead to the use of evolution to search the space of which script/heuristic is assigned to which unit. In this paper, we introduce the ideas of Portfolio Online Evolution and apply it to StarCraft micro, or individual battles. It is shown to outperform all other tested methods in battles of moderate to large size.