[Duke University] Development of an Autonomous Vehicle Testbed for Testing Intrusion Detection Systems

Kaustubh Sridhar, Miroslav Pajic

Poster.

Videos: (left) Shows lane keeping performed by eBuggy Autonomous Vehicle (right) Shows the onboard camera view and steps in the image processing methodology.

A software environment (in ROS (Robotic Operating System)) was created for the robot seen in the below video. Following which, a image processing pipeline was created such that the robot is able to stay within a lane using only its onboard camera. The pipeline consisted of identifying all edges, identifying those edges belonging to lane markers, classifying lane markers, using RANSAC to fit a polynomial and finding a polynomial to represent the lane. Further, few Intrusion Detection Systems were proposed and testing for the same was initiated.


[IISc Banglore] Bio-inspired Landing of Quadrotor Using Improved State Estimation

Hemjyoti Das, Kaustubh Sridhar, Radhakant Padhi

5th IFAC Conference on Advances in Control and Optimization Of Dynamical Systems (ACODS) 2018

Paper on ScienceDirect.

Video: Shows Parrot AR Drone landing along a bio-inspired trajectory at 6m height and 10 m away from landing zone

We present an improved state estimation technique - a fusion of Monocular SLAM (Simultaneous Localization and Mapping) and INS (Inertial Navigation System) and utilize it in landing a commercially available low cost quadrotor (Parrot AR Drone 2.0) in indoor environments along a trajectory generated by a bio-inspired guidance method.

[IIT Bombay] Trajectory Tracking using Backstepping Control on a Parrot AR Drone

Kaustubh Sridhar, Srikant Sukumar

Report.

Videos: (left) Parrot AR Drone tracking 8 shape trajectory (right) Parrot AR Drone tracking trajectory of RC Car

We implement backstepping control on a Parrot AR Drone 2.0 in simulation (Gazebo) and in hardware aided by a VICON motion capture system.