Neuro Inspired Adaptive Perception and Control for Agile Mobility of Autonomous Vehicles in Uncertain and Hostile Environments

Final Report: MURI: Neuro-Inspired Adaptive Perception and Control for Agile Mobility of Autonomous Vehicles in Uncertain and Hostile Environments Report Title This final report summarizes the results of the work performed between for the period beginning August 1, 2015 and ending July 31, 2016, under the support of ARO MURI grant no. W911NF1110046. During the last year, we made significant progress in several areas. First, we continued our investigation into the semi-autonomous and autonomous vehicle control, which has application to driver assistance scenarios (i.e., manned vehicles), teleoperated scenarios, or unmanned and autonomous vehicles. This year, a major focus was on conducting large-scale experimental analyses with human subjects in the loop, to study the engagement of humans with semi-autonomous and autonomous driving technologies (which were developed under this MURI program). We have also continued our work on improving the convergence rates of randomized, sampling-based planners, which have been recently shown to be capable for solving problems in high dimensional search spaces. We introduced three new algorithms, the PI-RRT# (that utilizes policy iteration updates), the DRRT (that combines gradient descent with randomized sampling to increase convergence) and the CL-RRT# (that uses closed-loop predictions for kinodynamic motion planning). We also investigated generalized label correcting (GLC) algorithms for kinodynamic motion planners and we found a very efficient scheme to generate, in a principled manner, the control primitives. In terms of perception, this last year we finalized the development of a new visual attention model which learns from human eye movements and continued our work on deciphering driver state and intentions beyond eye movements. Our perception work also focused on developing a SLAM-type of algorithm to support the MPPI controller described in last year's report. Last but not least, we continued our work on developing credible autocoders to simplify the validation and verification of autonomous embedded systems. This past year we have focused on autocoders for semi-definite programs, a very important class of on-line optimization algorithms that recently proved their worth with the autonomous landing of a SpaceX Falcon 9 rocket on a barge in the middle of the ocean. (a) Papers published in peer-reviewed journals (N/A for none) Enter List of papers submitted or published that acknowledge ARO support from the start of the project to the date of this printing. List the papers, including journal references, in the following categories: 02/08/2017 02/08/2017 02/08/2017 02/08/2017 09/02/2014 09/02/2014 09/02/2014 09/02/2014 09/06/2015 09/06/2015 09/06/2015 09/06/2015 09/06/2015 Received Paper 132 151 149 148 85 86 87 88 121 133 122 123 124 T. Wang, R. Jobredeaux, M. Pantel, P.-L. Garoche, E. Feron and D. Henrion. Credible Autocoding of Convex Optimization Algorithms, Journal of Optimization and Engineering, (08 2015): 0. doi: Park, J., Reimer, B. and Iagnemma, K.. A User Study of Semi-Autonomous and Autonomous Highway Driving: An Interactive Simulation Study, IEEE Pervasive Computing, ( ): . doi: Brian Paden, Valerio Varricchio, and Emilio Frazzoli. Design of Admissible Heuristics for Kinodynamic Motion Planning via Sum of Squares Programming, Robotics Letters, ( ): . doi: Emmanuel Boidot. Aude Marzuoli, and Eric Feron. Optimal players policies for discrete and continuous ambush games, IEEE Transactions on Intelligent Transportation Systems, ( ): . doi: Raghvendra V. Cowlagi, Panagiotis Tsiotras. Curvature-Bounded Traversability Analysis in Motion Planning for Mobile Robots, IEEE Transactions on Robotics, (08 2014): 1011. doi: 10.1109/TRO.2014.2315711 A. Borji, L. Itti. Defending Yarbus: Eye movements reveal observers' task, Journal of Vision, (03 2014): 1. doi: 10.1167/14.3.29 Laurent Itti, Ali Borji. Optimal attentional modulation of a neural population, Frontiers in Computational Neuroscience, (03 2014): 1. doi: 10.3389/fncom.2014.00034 D.-N. Ta, K. Ok, F. Dellaert. Vistas and parallel tracking and mapping with Wall–Floor Features: Enabling autonomous flight in man-made environments, Robotics and Autonomous Systems, (11 2014): 1657. doi: 10.1016/j.robot.2014.03.010 Junghee Park, Sisir Karumanchi, Karl Iagnemma. Homotopy-Based Divide-and-Conquer Strategy for Optimal Trajectory Planning via Mixed-Integer Programming, IEEE Transactions on Robotics, (12 2015): 0. doi: 10.1109/TRO.2015.2459373 Sterling J. Anderson, James M. Walker, Karl Iagnemma. Experimental Performance Analysis of a Homotopy-Based Shared Autonomy Framework, IEEE Transactions on Human-Machine Systems, (04 2014): 190. doi: 10.1109/TSMC.2014.2298383 A. Borji, D. Parks, L. Itti. Complementary effects of gaze direction and early saliency in guiding fixations during free viewing, Journal of Vision, (11 2014): 0. doi: 10.1167/14.13.3 Ali Borji, Laurent Itti, Daniel Parks. Augmented saliency model using automatic 3D head pose detection and learned gaze following in natural scenes, Vision Research, (11 2014): 0. doi: 10.1016/j.visres.2014.10.027 Ali Borji, Laurent Itti. Optimal Attentional Modulation of a Neural Population, Frontiers in Computational Neuroscience, (03 2014): 0. doi: Number of Papers published in peer-reviewed journals: Number of Papers published in non peer-reviewed journals: (b) Papers published in non-peer-reviewed journals (N/A for none) 1) Arslan, O. and Tsiotras, P., "Dynamic Programming Principles for Sampling-Based Motion Planners," Optimal Robot Motion Planning Workshop in the IEEE International Conference on Robotics and Automation, Seattle, WA, May 30, 2015. 2) Arslan, O. and Tsiotras, P., "Machine Learning and Dynamic Programming Algorithms for Motion Planning and Control," New England Machine Learning Day, Microsoft Corporation, Cambridge, MA, May 18, 2015. (c) Presentations 09/06/2015