ECE Rolling Robot Report

The Rolling Robot uses a unique form of locomotion to reach a designated goal. TERRA, as the rolling robot is affectionately called, can position itself in its environment utilizing an Ultra Wideband positioning system and plan a path to it goal. Once TERRA has reached its goal it will send back temperature data to its base station.

Credits

Electrical and Computer Engineering Team

Jack F Bell
Eric Elias
Alexander D Kleerup
Theodore M Leon

Mechanical Engineering Team

Jason Kahn
Cris Ramos
Ali Mganga
Vision Aryal

Mentors

Jack Mottley, Dan Phinney, & Christopher Muir

Abstract

Not all environments are reachable by humans, and therefore they require a specialized method of locomotion to explore them. The polyhedral robot has many advantages including but not limited to: omnidirectional movement, any orientationally state is stable and in normal operation, able to travel over rough terrain, not reliant on frictional forces, able to operate at relatively high speeds. The technology we plan to develop could be relevant towards many applications including space exploration and search and rescue operations. Our proposed design project is a rolling robot of a polyhedral shape. This form of movement brings advantages such as increased versatility in the terrain it can cross. Previously, rolling robots like the one our group envisions have manipulated its center of mass to invoke a rolling motion. We plan to deploy linear actuators to invoke a rolling motion. Potential deployments of the rolling robot, affectionately called TERRA, include disaster relief situations and planetary exploration. The rolling robot is exceptional at rolling over debris and rubble. The rolling robot could be deployed in a disaster relief situation being controlled by ultra wideband cellular towers and sending data about the condition of the terrain autonomously. Moreover, NASA brainstorms ideas for potential rovers for other planets that have new and novel designs. These designs utilize new ways to solve a problem that are specially adapted to a particular environment.

Design

Physics of a Rolling Robot

NX rolling model used to calculate minimum motor torque requirement

Circuitry

Circuit Block Diagram

Schematic

Electronics Panel

Component Selection, Hardware, and Budgeting

List of Materials

Raspberry Pi 4 4gb Model
12V 6Ah Battery
Adafruit AHT20 – Temperature & Humidity Sensor
Adafruit DPS310 Precision Barometric Pressure / Altitude Sensor
LSM6DS33 3D Accelerometer and Gyro Carrier with Voltage Regulator
SainSmart 16-Channel 12V Relay Module Board
Pololu 47:1 Metal Gearmotor 25Dx52L mm HP 12V
Pololu Dual MC33926 Motor Driver Carrier
Sparkfun SX1509 GPIO Breakout
SparkFun Pi Wedge 40-Pin
BOJACK 10SQ050 Schottky diode
12 & 20 Gauge Wire
Barrier Strips
3D printed parts
5V Voltage Regulator
Aluminum Sheet metal
5.5 Amp Fuse
PCB standoffs
Muffin Fans
LED Light Bar
PNP and NPN Transistors
1k Ohm Resistors
Micro USB Cables
Mechanical Limit Switches
Raspberry Pi Breakout board and Ribbon Cable
Raspberry Pi case and fan
JST Connectors
Tape
DWM1001-DEV Ultra Wideband Boards

Budgeting

Running Budget Spread Sheet of Rolling Robot Expenditures

Pie Chart of remaining budget. Total budget was 1,000 USD

Mechanical Engineer Update

Hardware Driver Code

A majority of the sensors and components used in our robot are meant to interface with a Raspberry Pi and have Python libraries already written for them. We are current working on implementing all of those libraries and testing that they work.

Ultra Wide Band Indoor Positioning System

Ultra wide band (UWB) is a radio technology we are using to make a positioning system. Our UWB System involves using four stationary boards as anchors and three boards attached to the robot as mobile tags. Both are used to create a system that communicates and tells the location of the tags in relation to the anchors.

We are using the DWM1001 Development Board from Decawave.

We have successfully interfaced between the Raspberry Pi and the UWB boards using a Python program that outputs X, Y, Z positions as well as the quality between 0-100. The tags are plugged into the Raspberry Pi’s USB ports and the anchors are setup using an Android application and Bluetooth.

Planning Algorithm

Finished planning algorithm
Robot moves in the direction that has the lowest angle to the goal
Problem: TERRA rolls in a grid. With only this planning law, the robot could oscillate back and forth and get stuck
Solution: Add a constraint that each roll cannot backtrack

TERRA rolling to a goal of (x: 1.5, y: 1,5).

Particle Filter

Extended Kalman filter Jacobian math was very cumbersome
- Required taking partial derivative of a quaternion with respect to another quaternion
Particle filter does not involve quaternions, only direct propagation
Each particle is a hypothesis of the robot’s position in the state space (x, y, z)
Finds the predicted position of UWB board for each particle

Particle Filter Algorithm

Propagate all particles forward (with noise) by the expected motion of the robot
Find the expected UWB measurement for each particle
Find the probability of the expected measurement in the context of the actual measurement
Resample particles proportional to their probability

Example

Simulation in Gazebo showing TERRA rolling towards a goal of (x: 1, y: 1) with the particle filter being visualized in RViz. During the roll, the filter stops to avoid the need for a more complex motion model. With each roll, noise is added to the particles. The distribution converges as the filter begins running again.

Mechanical Design Report