Krishna Varadarajan

These are some of fun projects that I have worked on in the past in the domain of robotics.

Autonomous Control of a RC Car (Code)

Remote controlled vehicles have the potential to go faster than it can usually be controlled. By converting this remote-controlled car to an autonomous car, powered by an Ardunio, this limitation on the performance can be removed. The goal of the project was introducing an Arduino-controlled RC type vehicle as a lab in the course curriculum of the senior-level Mechatronics course at Penn State, a demonstration on how an Arduino controlled RC car operates, and an example competition for the students.

Autonomous Object Picking Arm (Code)

Autonomous object detection and grasping has a wide range of application across various domains. The goal of this project was to design and develop an autonomous oject picking machine. The arm was mounted on a rotating disk with a range of 360 degree. The object detection module was built using a Pixy camera, and the actuation and control module was deployed on an Arduino. The device was capable of detecting a red object anywehre in its 360 degree range, pick it up and place it at a pre-assigned location. Inverse kinematics of a two-link arm was used to determine the initial position of the object.

Autonomous Maze Search and Rescue Robot

Autonomous robots can come to assist humans during various challenging situations, one such environment being search and rescue operations. Tennis balls were used as resuce targets inside a random maze. The robot used a fish eye camera to detect the ball's pixel location and orientation. 5 long and short-range IR sensors were used to navigate the maze. Navigation algorithm ran on an Arduino and the computer vision algorithm for detecting the tennis balls ran on a Raspberry Pi. The boards communicated through serial ports. The robot switched between various navigation modes and the ball-fetching mode using a finite state machine.

Benefts of Using Multiple Sensors on Reliable Space Headway Estimates (Paper Abstract)

Reliable space headways are critical for safe operation of connected and autonomous vehicles. These headway estimates also need to be accurately designated to the ego-vehicle's lane. The goal of the project was to understand the role of using multiple sensors in achieving the furthest reliable headway estimates of an open maneuvering space. It was observed that by using a vehicle-mounted camera along with a bumper-mounted RADAR and GPS mounted on a commercial truck, space headways can be detected upto and beyond the range of a RADAR.