Long-Term Occupancy Grid Prediction Using Recurrent Neural Networks

Abstract

In this work, we tackle the problem of modeling the vehicle environment as dynamic occupancy grid map in complex urban scenarios using recurrent neural networks. Dynamic occupancy grid maps represent the scene in a bird's eye view, where each grid cell contains the occupancy probability and the two dimensional velocity. As input data, our approach relies on measurement grid maps, which contain occupancy probabilities, generated with lidar measurements. Given this configuration, we propose a recurrent neural network architecture to predict a dynamic occupancy grid map, i.e. filtered occupancy and velocity of each cell, by using a sequence of measurement grid maps. Our network architecture contains convolutional long-short term memories in order to sequentially process the input, makes use of spatial context, and captures motion. In the evaluation, we quantify improvements in estimating the velocity of braking and turning vehicles compared to the state-of-the-art. Additionally, we demonstrate that our approach provides more consistent velocity estimates for dynamic objects, as well as, less erroneous velocity estimates in static area.

Deep Learning-Based Vehicle Behavior Prediction for Autonomous Driving Applications: A Review

“…The aforementioned works do not consider sensor impairment in the input representation. To overcome this drawback, a dynamic occupancy grid map (DOGMa [40]) is exploited in [41], [42]. DOGMa is created from the data fusion of a variety of sensors and provides a BEV image of the environment.…”

“…[41], [42] A top-down grid representation. Each cell contains the probability of the cell occupation, and its velocity.…”

“…steps. In [41], [42], the trajectory is predicted by estimating the vehicles occupancy likelihood for each cell in the dynamic occupancy grid map (DOGMa [40]) and each time-step in prediction horizon. They create DOGMa by assigning a grid map to a bird's eye view of the environment around the EV.…”

Dynamic Environment Prediction in Urban Scenes using Recurrent Representation Learning

“…This baseline was also used by Hoermann et al [7]. The assumptions made by this approach are: grid cell independence, linear, noisy particle dynamics, and absence of free space estimation (particles only capture occupied space) [7], [37]. Due to the last assumption, we use the predicted occupied mass values as a proxy for the occupied probabilities.…”

Short-Term Prediction and Multi-Camera Fusion on Semantic Grids

“…However, all of these work are in the sensor-dependent image space, which unnecessarily complicates the task considering downstream sensor fusion. In contrast, [6,7,19,20] and [24] employ an end-to-end trainable recurrent architecture to directly predict an unoccluded occupancy grid from laser data, capable to track multiple objects. We distinguish ourselves by investigating the capacity of our architecture in the context of sensor fusion and further, by using camera data instead of laser data, enabling semantically richer representations.…”