LiCaNext: Incorporating Sequential Range Residuals for Additional Advancement in Joint Perception and Motion Prediction

TransFuser: Imitation with Transformer-Based Sensor Fusion for Autonomous Driving

“…Forecasting: Most sensor fusion works consider perception tasks, e.g. object detection [14]- [16], [18]- [23], [47]- [60] and motion forecasting [24]- [30], [49], [61], [62]. They operate on multi-view LiDAR, e.g.…”

Exploiting Multi-Modal Fusion for Urban Autonomous Driving Using Latent Deep Reinforcement Learning

“…This chapter adopts sequential latent maximum entropy (MaxEnt) RL [32] Next, MaxEnt RL is trained on this latent space to learn a driving policy that ensures safety, efficiency, and comfort in an urban environment [172]. The perception and motion prediction are generated using LiCaNext [167], a novel real-time multi-modal feature fusion network that generates accurate pixel-wise joint perception and motion prediction. LiCaNext secured substantial accuracy gain over state-of-the-art networks and has proven its success in detecting small and distant objects.…”

“…A single LIDAR sensor is used to extract bird's-eye view (BEV), range view (RV), and residual input images. These images are passed into LiCaNext[167] (Chapter 5), a real-time multi-modal fusion network, to generate accurate joint perception and motion prediction. Next, predictions are fed with another simple BEV image into the latent DRL to learn a complex end-toend driving policy ensuring safety, efficiency, and comfort.…”