Dynamic vision sensors have been investigated to report motion-only images for moving object recognition. Less but essential information helps post-process recognition algorithm reduces computationimproves accuracy. Implement low powerlow latency deep Learning chip based on neuromorphic Intelligence. The neuromorphic obstacle detection algorithm integrates visualproprioceptive signals. The algorithm is characterized by its efficiencylow power consumption. We possess the next-generation in-memory computing AI chipsnext-generation UAV key softwarehardware technology

Operated at the lowest supply 0.8V with ping-pong PWM pixel to present the 1st multi-mode vision sensor featuring 71.2uW 360fps image capturing for object recognition, 74.4uW 510fps full-resolution frame-based event reporting for motion detection,121.6uW 890fps block-level saliency detection for object tracking. The sparse algorithm is to store non-zero weights on the CIM can save inference timetotal energy consumption, also reduce the number of parameters by 2048 times. Fusing optical flow, SNN,proprioception on UAV to perform obstacle avoidance in complex environment.

The dynamic vision sensor has the characteristics of low power consumptiondetecting moving objectscombining stereo vision computing to achieve the distance detection of high-speed moving objects  to use in security monitoringUAV obstacle avoidance. With a model compression framework of implementing pruning algorithm on SRAM-CIM hardware accelerator, it can be applied on the IoT device, realize low powersmart computing. In-memory computing AI chips is used in medical treatment, robotics,smart home appliances. UAVs is for inspection, agriculture,transportation.

天使投資人、策略合作夥伴