A Hierarchical Network for Diverse Trajectory Proposals Sriram N. N. 1 , Gourav Kumar 1 , Abhay Singh 1 , M. Siva Karthik 2 , Saket Saurav 1 Brojeshwar Bhowmick 3 and K. Madhava Krishna 1 Abstract— Autonomous explorative robots frequently en- counter scenarios where multiple future trajectories can be pursued. Often these are cases with multiple paths around an obstacle or trajectory options towards various frontiers. Humans in such situations can inherently perceive and rea- son about the surrounding environment to identify several possibilities of either manoeuvring around the obstacles or moving towards various frontiers. In this work, we propose a 2 stage Convolutional Neural Network architecture which mimics such an ability to map the perceived surroundings to multiple trajectories that a robot can choose to traverse. The first stage is a Trajectory Proposal Network which suggests diverse regions in the environment which can be occupied in the future. The second stage is a Trajectory Sampling network which provides a finegrained trajectory over the regions proposed by Trajectory Proposal Network. We evaluate our framework in diverse and complicated real life settings. For the outdoor case, we use the KITTI dataset and our own outdoor driving dataset. In the indoor setting, we use an autonomous drone to navigate various scenarios and also a ground robot which can explore the environment using the trajectories proposed by our framework. Our experiments suggest that the framework is able to develop a semantic understanding of the obstacles, open regions and identify diverse trajectories that a robot can traverse. Our comparisons portray the performance gain of the proposed architecture over a diverse set of methods against which it is compared. I. I NTRODUCTION Autonomous navigation requires the explorative ability to navigate by identifying diverse paths to multiple goal points by perceiving its environment. A simple instance is a case where an autonomous drone using SLAM [1] or using any other sensor based reconstruction [2], [3] can manoeuvre between obstacles by going around towards their right or left side. Similarly, in an indoor corridor intersection, it can proceed straight, right or left. The autonomous robot has to identify various possible paths and goal points that it can pursue in any given setup. Such a task is easy for humans where they can map the scene configuration to identify multiple traversable areas, goal points and also a fine grained path. One such use case in autonomous vehicle setting is where you know the rough direction of travel but not the actual goal point. A diverse path prediction is required in this case to determine the best plan of action. Some of the key applications can include but not limited to: • In case of rerouting alternate trajectories might be required due to sudden unexpected changes in the environment. 1 International Institute of Information Technology, Hyderabad, India 2 University of Heidelberg, Germany 3 TCS Research and Innovation Labs, Kolkata, India Fig. 1. Trajectory Proposal:(a) Bird’s-eye-view of the pointcloud data generated at a road bifurcation. This pointcloud data is converted into a 2D occupancy grid and passed as an input to TPNet (b) Proposed trajectories in the form of probability values for each pixel which is passed to TSNet (c) Sampled trajectory output. (d) Third-person-view of the driving scenario illustrating proposed trajectories with respect to the car. • It also opens up the possibility of reaching the destina- tion in multiple possible ways. • It can be useful in overcoming GPS errors. The net- works output can help in guiding the vehicle precisely to take turns thus overcoming the state estimation errors that results in early or late turns near an intersection as shown in figure 2. In this work, we present a Convolutional Neural Network framework which precludes the need for explicit determi- nation of candidate waypoints as it learns a direct mapping from intermediate semantic representation to candidate tra- jectories as shown in fig. I. Our framework which consists of two stages. The first is a Trajectory Proposal Network (TPNet), an encoder-decoder style network, which uses the occupancy map and robot’s past trajectory as input and produces multiple traversable areas by inherently discovering the waypoints possible in the scene. This is achieved through our novel supervision based on multiple choice learning which encourages the TPNet to identify various waypoints and propose diverse traversable areas for a given scenario. The second stage of our framework, the Trajectory Sampler Network (TSNet) is a Long Short-Term Memory (LSTM) based network which uses the proposals by TPNet and sam- ples precise trajectories values through those proposals which can lead the robot to various goal points. This alleviates the arXiv:1906.03584v1 [cs.RO] 9 Jun 2019