Summary for 2026-04: Delivered three core capabilities enabling realistic simulation-based validation and safer autonomous behavior in Carla-based tests. Key business value includes faster iteration, higher fidelity testing, and lower integration risk in CI. - OpenCV-compatible YOLOv3-tiny integration: real-time inference compatibility and reliable project execution within the existing pipeline. - Carla simulation environment enhancements: NPC vehicle generation, third-person automatic following of the main vehicle, vehicle bounding box displays, and the ability to spawn obstacles alongside vehicles, with upstream main merges to maintain alignment and resolve conflicts. - Vehicle obstacle detection, tracking, and avoidance enhancements: obstacle recognition with automatic braking, collision sensing, intelligent avoidance; DeepSORT-based tracking with Hungarian algorithm; TTC-based safe-distance calculation and optimized lane-change triggering; collision sensors added. Overall, these changes improve test realism and coverage, reduce integration risk, and accelerate validation cycles for perception, planning, and control stacks.