1 of 1

WINLAB

Daniel Mahany (HS), Dhruv Ramaswamy (HS), Nandini Venkatesh (HS) - 2024

Remotely Piloted Vehicles Project

Project Overview

Low latency networking is an emerging technology for use in remote sensing and control of vehicles and robots. Our motivation was to evaluate the usability of remote piloting over an internet connection.

Main Objectives:

Software & Networking:

Enable wireless control of each vehicle and extend the operational range using a VPN. Develop an intuitive and user-friendly interface and establish a reliable and low-latency camera feed.

Hardware:

Build two vehicles with distinct hardware and features: one equipped with mecanum drive and the other with Ackermann steering. Integrate additional sensors, emergency stop mechanisms, and other features.

Testing and Driver Preferences

Turn #1 in testing course

Turn #2 in testing course

Experimental Setup

Results of the Experiment:

Future Work:

Network Architecture

“Server”

VPN (ZeroTier)

Direct Connection

Camera Data

Movement Commands

“Client”

Hardware

The use of mecanum wheels allow for both “tank” drive_(a) as well as omnidirectional movement. We initially hypothesized that the ability to strafe left and right_(b) would allow the pilot to better navigate around corners and turns. Additionally, this drive train allows for diagonal movement_(c).

Client Software - UI

Connection Delay in milliseconds

Mecanum Virtual Controls (emulated by keypress)

Camera Feed & Reload button

Mecanum User Interface

Connection Delay in milliseconds

Ackermann Virtual Controls (emulated by keypress)

Servo Rotation

Camera Feed & Reload button

Ackermann User Interface

Ackermann Steering:

This drive train features steering similar to a traditional road vehicle. Our vehicle is modified from the OSOYOO Servo Steer Smart Car kit.

Server Software

Emergency Stop

Gathers the distance data 100 times each second. If triggered it, roughly stops an inch from the obstacle. This feature can be toggled on and off.

Troubleshooting

Experiment with different image encoding and decoding to decrease camera latency.
Improve mecanum movement

Calculate the amount of energy to send to each motor according to joystick vector
Resolve weight and balance issues to reduce drift.

Retrieve sound data from the webcam microphone.
Potentially improve safety system by adding depth sensing and more sensors for emergency stops.

	Trial 1	Trial 2
Ackermann Vehicle	56.87 Seconds	42.43 Seconds
Mecanum Vehicle	55 Seconds	49.144 Seconds

Average Time to Complete the Course:

	Trial 1	Trial 2
Ackermann Vehicle	0.4	~ 0.3
Mecanum Vehicle	~ 0.5	0.4

Average Number of Collisions while Driving:

Mecanum Drive