In this paper we present a trajectory generation method for autonomous overtaking of static obstacles in a dynamic urban environment. In these settings, blind spots can arise from perception limitations. For example, the autonomous car may have to move slightly into the opposite lane in order to cleanly see in front of a car ahead. Once it has gathered enough information about the road ahead, then the autonomous car can safely overtake. We generate safe trajectories by solving, in real-time, a non-linear constrained optimization, formulated as a Receding Horizon planner. The planner is guided by a high-level state machine, which determines when the overtake maneuver should begin. Our main contribution is a method that can maximize visibility, prioritizes safety and respects the boundaries of the road while executing the maneuver. We present experimental results in simulation with data collected during real driving.

Abstract-We present the development of a full-scale "parallel autonomy" research platform including software and hardware. In the parallel autonomy paradigm, the control of the vehicle is shared; the human is still in control of the vehicle, but the autonomy system is always running in the background to prevent accidents. Our holistic approach includes: (1) a driveby-wire conversion method only based on reverse engineering, (2) mounting of relatively inexpensive sensors onto the vehicle, (3) implementation of a localization and mapping system, (4) obstacle detection and (5) a shared controller as well as (6) integration with an advanced autonomy simulation system (Drake) for rapid development and testing. The system can operate in three modes: (a) manual driving, (b) full autonomy, where the system is in complete control of the vehicle and (c) parallel autonomy, where the shared controller is implemented. We present results from extensive testing of a full-scale vehicle on closed tracks that demonstrate these capabilities.

In this paper, we introduce a novel method to interpret recurrent neural networks (RNNs), particularly long short-term memory networks (LSTMs) at the cellular level. We propose a systematic pipeline for interpreting individual hidden state dynamics within the network using response characterization methods. The ranked contribution of individual cells to the network's output is computed by analyzing a set of interpretable metrics of their decoupled step and sinusoidal responses. As a result, our method is able to uniquely identify neurons with insightful dynamics, quantify relationships between dynamical properties and test accuracy through ablation analysis, and interpret the impact of network capacity on a network's dynamical distribution. Finally, we demonstrate generalizability and scalability of our method by evaluating a series of different benchmark sequential datasets.

Moving obstacles occluded by corners are a potential source for collisions in mobile robotics applications such as autonomous vehicles. In this paper, we address the problem of anticipating such collisions by proposing a vision-based detection algorithm for obstacles which are outside of a vehicle's direct line of sight. Our method detects shadows of obstacles hidden around corners and automatically classifies these unseen obstacles as "dynamic" or "static". We evaluate our proposed detection algorithm on real-world corners and a large variety of simulated environments to assess generalizability in different challenging surface and lighting conditions. The mean classification accuracy on simulated data is around 80% and on realworld corners approximately 70%. Additionally, we integrate our detection system on a full-scale autonomous wheelchair and demonstrate its feasibility as an additional safety mechanism through real-world experiments. We release our real-timecapable implementation of the proposed ShadowCam algorithm and the dataset containing simulated and real-world data under an open-source license.

Today, head-up displays (HUDs) are commonly used in cars to show basic driving information in the visual field of the viewer. This allows information to be perceived in a quick and easy to understand manner. With advances in technology, HUDs will allow richer information to be conveyed to the driver by exploiting the third dimension. We envision a stereoscopic HUD for displaying content in 3D space. This requires an understanding of how parallaxes impact the user's performance and comfort, which is the focus of this work. In two user studies, involving 49 participants, we (a) gather insights into how projection distances and stereoscopic visualizations influence the comfort zone and (b) the depth judgment of the user. The results show that with larger projection distances both the comfort zone and the minimum comfortable viewing distance increase. Higher distances between the viewer and a real world object to be judged decrease the judgment accuracy.