Driving in India poses unique challenges due to road debris, stray animals, pedestrians, uneven surfaces, and hazards like dooring — where an opening car door strikes passing cyclists. Existing sensing technologies have notable limitations: radar lacks spatial resolution, LiDAR is too costly, and conventional cameras underperform in low light. Time-of-Flight (ToF) cameras offer a practical alternative, combining active illumination with per-pixel depth measurement for accurate ranging up to ~6 metres. This work investigates ToF sensors for parking assistance, dooring prevention, and slow-traffic manoeuvring, supported by a deep neural network for object detection and classification.

• Understanding limitations of existing sensing technologies in identifying road safety hazards specific to India
• Demonstrate how ToF cameras can be deployed for parking assistance, dooring prevention, and slow-traffic manoeuvring.
• DNN models for reliable object localization and cloassification under poor-illumination and considering the noise in ToF cameras.