Predictive ADAS in a Digital Twin: Overtaking with 2 second Traffic Forecast, Exit & Red-Light Stop.

This video demonstrates a camera-first Advanced Driver Assistance System (ADAS) driving in closed loop in a simulated environment. The system combines camera perception, HD-map guidance, radar and LiDAR confirmation, and motion prediction of surrounding vehicles, and controls steering, throttle and braking in real time. The test environment serves as a virtual proving ground in the spirit of a digital twin: the complete perception, planning and control software runs unchanged against simulated camera, radar and LiDAR data from a sensor configuration modelled on production ADAS hardware. This software-in-the-loop setup allows scenarios such as overtaking, highway exits and red-light stops to be reproduced exactly, measured against ground truth and repeated after every change — before any testing on real roads. The drive starts on a multi-lane highway with mixed-speed traffic. The system follows the lead vehicle at a safe gap, plans and executes overtaking lane changes, and warns of fast vehicles approaching from behind before moving over. It then changes lanes for a highway exit, leaves the highway, turns through a junction and enters urban traffic, where it recognises the governing traffic light, queues behind other vehicles and stops at the line on red. The right-hand panel shows a surround bird's-eye view with the planned path, surrounding vehicles and lane-change intent, together with LiDAR coverage, radar returns, and predicted vehicle positions at +1 second and +2 seconds. 本影片展示一套以攝影機為主的先進駕駛輔助系統（ADAS），於模擬環境中進行閉迴路駕駛。系統整合攝影機感知、高精地圖導引、雷達與光達確認，以及周圍車輛之運動預測，並即時控制轉向、油門與煞車。 本測試環境以數位孿生（Digital Twin）的概念作為虛擬試驗場：完整的感知、規劃與控制軟體在不作任何修改的情況下，直接以模擬之攝影機、雷達與光達資料運作，感測器配置參考量產 ADAS 硬體設計。此軟體在環（Software-in-the-Loop）架構可精確重現超車、下交流道及紅燈停車等情境，與真值（Ground Truth）比對量測，並於每次修改後重複驗證，再進行實際道路測試。 行駛過程由多車道高速公路開始，車流速度不一。系統以安全距離跟隨前車，規劃並執行超車變換車道，並在變換車道前偵測後方快速接近之車輛並發出警示。接著系統為下交流道變換車道、駛離高速公路、通過路口轉彎並進入市區車流，辨識管制號誌、於車陣中排隊，並在紅燈時停於停止線前。 右側面板顯示環景鳥瞰圖，包含規劃路徑、周圍車輛與變換車道意圖，以及光達覆蓋範圍、雷達偵測結果，與周圍車輛於 +1 秒及 +2 秒之預測位置。 Test Features Lane detection and map-guided lane keeping Lead vehicle tracking and adaptive cruise Overtaking lane change (planned and executed) Rear fast-vehicle warning Highway exit and route following Junction turning Traffic light recognition and stopping Queue following in urban traffic Surround view and motion prediction (+1 s / +2 s) Simulation only — recorded in a simulated environment. Vehicle control shown here is performed in simulation and has not been tested on public roads. Performance may differ under real-world sensing, traffic and environmental conditions. 僅為模擬結果 — 本影片於模擬環境中錄製，影片中之車輛控制僅於模擬環境中執行，未於實際道路上測試。系統於真實感測、交通及環境條件下之表現可能有所不同。 This video is intended for autonomous driving research, ADAS development, system evaluation and technical demonstration purposes. 本影片僅供自動駕駛研究、ADAS 系統開發、系統評估及技術展示使用。 Website: https://sites.google.com/view/chingwe... E-mail: cwwang1979@gmail.com Copyright: Prof. Ching Wei Wang #AutonomousDriving #ADAS #DigitalTwin #VirtualTesting #SoftwareInTheLoop #Simulation #ClosedLoop #HighwayDriving #UrbanDriving #Overtaking #LaneChange #AdaptiveCruiseControl #LaneDetection #LaneKeeping #VehicleTracking #TrafficLightRecognition #MotionPrediction #SensorFusion #BirdsEyeView #ComputerVision #SelfDrivingCar