Fig. 5: Robot obstacle avoidance system driven by neuron firing features.

A Schematic diagram of the robot avoiding an obstacle that appears at different distances. B The angular and linear velocities are linearly proportional to the BDN frequency and SDN frequency, respectively. A reasonable maximum linear velocity (0.9776 m/s) is set here. C An illustration of the relationship between variable distance range and the responses determined by SCNC, involving angular velocity and linear velocity. D The firing patterns of the MCN (blue), BDN (yellow) and SDN (red) are associated with the ‘distance coded’ signal, ‘steering’ signal and ‘actuating’ signal, respectively. E The indoor trajectory recording of a robot vehicle under three typical features. The motion trace is in the cyan line, and the steering angle is in yellow. F Comparison of time latency between the obstacle avoidance scheme implemented by conventional methods (NVIDIA Jetson AGX Xavier), our physical SCNC and the SCNC after scaling. And a comparison of power consumption between the obstacle avoidance scheme implemented by FPGA, our physical SCNC and the SCNC after scaling.