Autonomous, Self-Balancing Bicycle for Urban delivery and Community Connection

Overview

Background Since Knight Rider in 1982, the prospect of autonomous vehicles has lingered in the background for over forty years. It makes sense why so many leading firms are racing to perfect and implement this still new technology. Automating something eliminates any possibility of human error, and performs tasks more quickly and more efficiently. In this project, the end goal is to develop a fully autonomous bicycle, capable of riderless driving and self-balancing on two wheels. By creatively applying AI, object recognition, and PID control, in addition to an effective balancing mechanism, I would like to uniquely implement these aspects through this project. Problem and Solution Delivery can be a time-consuming experience, both for the driver and for the recipient. Urban cities are often plagued with traffic congestion and uneven, unmarked roads, slowing down delivery by car. Tight alleyways further hinder access. Human error is often a contributing factor that furthers possibility for error. As such, many delivery companies like UberEats have resorted to delivery by bicycle to avoid the inefficiencies of traffic. Automating delivery has also become more and more appealing in its efficiency, and technologies in this bicycle can also be applied to other autonomous vehicles and AI systems. My goal is to apply Autonomous technology–including image recognition and PID feedback control–in the form of a fully autonomous, self-balancing bicycle. The bicycle is compact without extra wheels, and makes delivery both an efficient and safe experience. Approach My approach to this project involves the breakdown of general project goals to address problems, followed by a specific method of implementation. The first main stage of the project (autonomous driving) is pedestrian and road safety, which includes a few planning steps. First, utilize software to detect pedestrian(s), second, determine the bicycle’s distance from pedestrian(s), and third, use sensor fusion to initiate bicycle response. These features can be then implemented with a camera module and AI object recognition to detect the pedestrian, a proximity sensor to determine distance from pedestrians, and finally the sending of a command controlling motors to stop the bicycle. The second main stage of this project is self-balancing. The general project goals here are more straightforward. First, determine the bicycle’s axis of tilt with respect to gravity, and second, design a mechanism capable of physically responding to correct any tilt. These can be implemented with an IMU sensor (inertial sensor and gyroscope) which measures the roll axis, a feedback loop that will analyze and initiate a response, and a flywheel assembly to provide the torque and angular momentum needed to balance the bike. Conclusion, results, and future work Using this systematic approach, I have been able to implement full driving functions and a software library– a set of APIs for bike control, currently containing seven functions for braking, driving, and other bike functions. Additionally, in concentrating on autonomous driving, I’ve implemented AI techniques including image processing, object recognition, automatic braking, and obstacle avoidance to facilitate autonomous driving and safety. Additionally, through thorough analysis of collected data and experiments– including weight factors, battery weight, various battery classes, monitoring PWM, and analyzing current draw of the control and drive motors– I have determined the optimal battery capacity and type to maximize range and runtime, and minimize weight. Using these results, I have drafted a detailed plan to continue work on this project, including enhancements to the battery efficiency and range of the bicycle, as well as implementation of the self-balancing mechanism.

Awards (1)

• Category Award: HM

Competition history

• CSEF 2023 Electronics & Electromagnetics · Entry S1006

Resources

• CSEF published award list ↗
• CSEF published award list ↗
• CSEF published award list ↗
• CSEF public showcase (Internet Archive capture) ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Physics and Astronomy projects
• CSEF Physics and Astronomy projects
• Every CSEF 2023 project