Plenary Lecture
Bio-Inspired Visual Information Processing and its Applications to DSM and ADAS
Professor I. S. Han
Graduate School for Green Transportation
Korea Advanced Institute of Science and Technology
Korea
Email: i.s.han@kaist.ac.kr
Abstract: The physiological studies about visual cortex from the investigation of cat’s striate cortex by Hubel and Wiesel have introduced the foundation knowledge about biological vision in the nature. The neuromorphic visual information processing method, inspired by Hubel and Wiesel’s experiments, is proposed to replicate the performance of visual cortex in practical computing settings. By applying the orientation feature extraction and subsequently applying the neural network ensured robustness and accuracy. Considering the number of fatalities and serious injuries of road users, the safety enhancement has begun to gain more attention, in particular accurate and timely detection of the risk of accident with the focus on the drivers and other vulnerable road users, such as the pedestrians or cyclists. We have proposed that the neuromorphic visual processing algorithm based on the biological vision system is an effective approach for making detection of human figures from a moving vehicle. The effectiveness of neuromorphic vision has been evaluated for the vulnerable road users of cyclist or pedestrian detection via successful detection at either the day time or the night time. The post enhancement with deep networks shows that further applications of incorporating neuromorphic visual processing into Driver State Monitoring both for the purpose of enhancing vulnerable road users’ safety and the extended human-machine interface of emotive detection. The early implementations have demonstrated the advantages of fast and robust neuromorphic vision with either the small embedded system based on Raspberry or the customized embedded system based on FPGA, while the neuromorphic ASIC based on Hodgkin-Huxley formalism was evaluated successfully based on the controlled CMOS conductance.
Our conclusion is that the neuromorphic vision mimicking of the visual cortex, coupled with neural networks, suggests it as the new smart and robust device of Drive State Monitoring and Advanced Driver Assistance System for the road safety enhancement.
Brief Biography of the Speaker: Prof. Il Song Han is an invited professor at the Graduate School for Green Transportation at Korea Advanced Institute of Science and Technology, and has significant experience in academia and industry. His research interests are in the areas of analogue-mixed VLSI design, wireless power transfer, neuromorphic device and vision system, bio-inspired neural networks VLSI, active safety technology for vulnerable road users and driver status monitoring. One of his neuromorphic vision research results has been adopted for the new safety device development by an automotive manufacturing industry.