Solution Development

Solution:

• Proposed Solution: Follow-ME Robot

• Requirements:

  • Object-Detection

  • Path-finding

  • Self-Driven

  • Picks up and puts down objects

  • A user interface (UI) to control the robot's activities

Design Tasks:

• Start by getting OpenCV and a TensorFlow model running in Python to detect objects.

• Migrate this image-detection model to the NVIDIA Jetson Nano board.

• Create code that will allow the robot to move using inverse kinematics.

• Fuse the object-detection code with the robot's inverse kinematics code and gather distance data from the light detection and ranging (LiDAR) sensor to allow for movement towards objects.

• Add a robotic arm that then will allow the robot to pick up and manipulate objects in its surroundings.

Design Goals:

• User-friendly interface

• Should be cost-effective to install and operate

• Easily integrate into current working environments (hospitals, military bases, etc.)

• Easily trainable artificial intelligence (AI) models for various applications.

  • Specific Design Goals

    • Create a UI that allows users to program precisely where the robot is going and what it is supposed to be picking up.

    • High power efficiency to operate for long periods of time

    • Strong structure to endure rough environments

Design Components:

Electronics:

• NVIDIA Jetson Nano board to run our AI model

• LiDAR sensor to measure the distance to objects

• Stereographic camera for both distance and object detection

• Servo motors to control the manipulator

• H-bridge motor driver

• Microcontroller to control peripheries

Frame:

• Strong and durable frame

• Robotic arm/object manipulator

• Sensor mounts

• Wheels

• Motor mounts

• Standoffs

Software:

• OpenCV: Camera data parsing and processing

• TensorFlow: Neural network development for object detection model

• Inverse kinematics code to control motor drivers and extremities

Path Elimination:

• The idea of just solely using a camera or a LiDAR system to help detect objects and surroundings was eliminated because the robot would need the functionality of both to be able to work effectively and efficiently.

• We decided to eliminate the idea of using regular directional wheels in favor of using omnidirectional wheels due to their increased efficiency in regards to physically moving around complicated terrain.