Project
Report -
Videos
Skills - ROS, Python, Gazebo, Navigation, Object Recognition
PanduBot - A Personal Home Robot
Codebase -
 3d model of PanduBot |  PanduBot |  Interacting with humans |
|---|---|---|
 Taking customer order |  Map of 4th floor of Penn Engineering |
Phase-1: Simulation and 3D modeling
Indoor robots are gradually penetrating into our houses. Even though a lot of challenges with navigation in a cluttered environment have been solved, it still remains a difficult task to develop indoor robots. In this project, my teammates (Bhavya Gupta, Nitin Sanket, Paritosh Kelkar and Sarath Bharathi) and I develop a robot capable of navigating indoors, identify various objects and interacting with humans. In the first phase of the project, we simulate the turtlebot 2 on ROS - gazebo and test for navigation and exploration algorithms. Also, a 3D model of the robot is developed in order to get its parts fabricated. The model is shown below.
Phase-2: Implementing vision and navigation algorithms on the robot
Post the simulation phase, the next task was to get the robot up and running. Various vision and navigation (AMCL) algorithms were implemented to perform high level task like point to point navigation and object recognition.
Phase-3: Service robot task
The final goal of the project was to develop a service robot capable of catering in restaurants. The videos above show the robot in action. For this task, I was jointly responsible for mapping the environment and recognition of objects.