The Empathetic Car: Exploring Emotion Inference via Driver Behaviour and Traffic Context
summary
Modified 10 cars collect CAN-bus signals and front‑view video over 700 hours of real-world driving and label emotion by aggregating a facial‑expression detector into 5‑minute segments. Recurrent neural nets for CAN‑only, video‑only and fusion are evaluated under personalized and LOSO splits. Fusion wins slightly; vehicle‑data models notably outperform Firstbeat HRV physiological baselines. The surprising claim is that non‑intrusive CAN plus traffic context can match medical‑grade sensors. Built system works in personalized settings but generalisability (LOSO) and model interpretability remain unresolved.
AI-written summary · about
related talks
10mVEmotion: Using Driving Context for Indirect Emotion Prediction in Real-Time
11mAmbient Intelligence and Emotionally Intelligent Devices
52mPhysical AI: Crafting Resilient Systems with Emotional Intelligence
15mRecognizing Emotional Expression in Game Streams
34mEnhancing Customer Experiences with Emotional AI,
49mAffective Computing – Bringing Humans and Machines Closer Through Emotions
19mWhat Can Be Predicted from Six Seconds of Driver Glances?
08mHuman–AI Collab for Reliable Automotive Control Systems