RAMP-CNN: A Novel Neural Network for Enhanced Automotive Radar Object Recognition

Millimeter-wave (mmW) radars are being increasingly integrated into commercial vehicles to support new advanced driver-assistance systems (ADAS) by enabling robust and high-performance object detection, localization, as well recognition - a key component of environmental perception. In this paper, we propose novel radar multiple-perspectives convolutional neural network (RAMP-CNN) that extracts the location class objects based on further processing range-velocity-angle (RVA) heatmap sequences. To bypass complexity 4D networks (NN), combine several lower-dimension NN models within our RAMP-CNN model nonetheless approaches performance upper-bound with lower complexity. The extensive experiments show proposed achieves better average recall (AR) precision (AP) than prior works in all testing scenarios (see Table. III). Besides, is validated work robustly under nighttime, which enables low-cost potential substitute for pure optical sensing severe conditions.