Intelligent Ground Robots
During my undergrad at University of Central Florida, I spent four years in the Robotics Lab working on autonomous ground robots where I got hands-on experience with many robotics topics, including computer vision, machine learning, electronics, embedded systems, trouble-shooting, debugging, wxWidget GUI displays, microcontroller work, sensor integration (GPS+vision+LIDAR), scripting languages, and mechanical work.
I worked on three robots: Black Knight, Calculon, and Gamblore that competed in the annual International Ground Vehicle Competition where several times we placed 3rd in a group of ~20 international teams. By the time I left, I was team captain of our robot and lead software developer for our 100,000 lines of C+ code library named Zebulon, about 35% of which I wrote myself. We practiced good software engineering and held ourselves to strict standards in coding conventions, documentation, and testing. In fact, years later I am still using parts of it in my projects.
Gamblore: To Ride a Robot
I demonstrate the robustness of our new custom-built Gamblore LX500 ground robot by riding it, cutting donuts in the parking lot, driving backwards around cones, and chasing birds at 7 am in the morning – all after staying up all night CNCing the axle and motor assembly.
Calculon: Vision System & Path Planning
This is a video feed of the GUI output of our Calculon robot. During autonomous navigation, it shows what the artificial intelligence is seeing and planning. It shows the raw camera image, the results of object recognition, the local map of obstacles (using vision + SICK laser), and also the planned path.
Calculon: Autonomous Challenge
At the International Ground Vehicle Competition (IGVC2005), Calculon took 4th place overall and 3rd place in design. This video shows our performance, speed up four times, in the autonomous challenge. The goal of the robot is to navigate the path designated by the white lines and while avoiding obstacles.
Calculon: First Vision System
This video made at 1 AM on May 22, 2005 shows my full vision system working for Calculon. It is identifying lines on the grass, construction barrels, white buckets, and potholes. On a P4 Alienware laptop, I was averaging 13 fps at 360×240. It is shocking to think how far we’ve come in computational power.