1 of 2

AutoChargerX

Team A1: Bruce Cheng, Steven Zhang, Harry Huang

18-500 Capstone Design, Fall 2024

Electrical and Computer Engineering Department

Carnegie Mellon University

System Architecture

Product Pitch

There are two main approaches for people to charge their electronic devices: wired charging and wireless charging. However, wired charging forces people to stay close with their devices and wireless charging requires perfect alignment for charging thus having a high failure rate.

To address the problems mentioned above, our project, AutoChargerX, introduces a new way in wireless charging technology, addressing the frustrations of wired and misaligned wireless chargers. By utilizing a 3-axis stepper motor gantry system, an underdesk device detection and control feedback computer vision system, monitoring softwares, and a sandwich glass table top with 4 wireless charging pads, our smart charging table automatically detects and aligns with electronic devices such as your cell phone, allowing seamless wireless charging for multiple devices without manual intervention.

Overall, the system has a ~93.3% electronic device detection accuracy within 1 second. The charging pad - electronic devices automatch time is ~5.47 seconds on average and the charging starts ~0.8s on average after successful alignment.

System Description

System Evaluation

Conclusions & Additional Information

System Assessment

The system delivers automated device charging through intelligent detection and integrated software solutions. It identifies compatible devices, manages charging pads, and enables real-time monitoring through iOS, macOS, and web applications, allowing users to track multiple devices charging simultaneously via their preferred interface.

Takeaways

  • Alternative approaches using existing components
  • Rapid prototyping/component selection techniques
  • Troubleshooting and debugging from SW, HW, Meche aspects

Future Development

  • PCB and commercial level integration of the system
  • Appearance design and various table size

Metric

Target

Actual

Auto Device Detection

<=1sec, <=1cm

Success Rate 93.33%

Auto Match and Alignment

<= 10sec, <= 1cm

Avg Alignment Error: 4.76mm

Avg Alignment Time: 5.47sec

Charging Start

<=2sec after alignment

Avg Charging start time: 1.4s

Fully Charge Time

<=45min

Completed

Phone Live Battery Display

<= 500ms Update rate

Completed

Temperature Control Monitor

Threshold at 50℃

Completed

Use-Case Requirements

Gantry Subsystem

It use timing belt and 3 motors to achieve the manipulator precise movement in 3-axis. The manipulator has four magnetics tips which will be used to attached to the glass desktop and move the charging pads inside.

Charging Pads

The wireless charging modules has a LED light on the PCB for charging status indication. The 3D-printed case with magnetics tips for attaching to gantry manipulator.

Vision Subsystem

It features a wide-angle camera under table. By utilizing YOLOv5 model that is running on cloud, it could detect devices on tabletop and send location command to gantry system�for charging pad movement.�It also monitor lights under�the charging pads which �serves as a indicator for�success alignment.

Software System

Y-Axis

X-Axis

Z-Axis

Motor

Manipulator

Figure 5: Phone App,Web App, and MacOS App (from left to right)

The system leverages a Raspberry Pi 5 as its central controller, utilizing GPIO pins to generate PWM signals for precise motor driver and motor control. A wide-angle camera captures continuous image frames, which are transmitted to a cloud VM for real-time phone detection using YOLOv5 machine learning model, and then detected coordinates are send back to Raspberry Pi. The platform also features a comprehensive device management system, which allows users to monitor multiple devices simultaneously. Upon installing the mobile application, user data synchronizes with cloud storage, enabling device tracking through either a macOS application or web interface.

https://course.ece.cmu.edu/~ece500/projects/f24-teama1/

More Info?

Scan me!

Design Requirements (30 Tests)

Sandwiched tabletop

\w 4 charging modules

Charging Initiation: Automatically align charging pad with user’s device

Fast detection/ Accurate alignment/ Quick Response

During & After Charging: Device monitoring with software and device APIs

Status Tracking/ Overheating Prevention

Finish charging fast/ Automatically stop

Figure 1. Block Diagram for system architecture

Figure 2. Gantry Subsystem Mechanism

Figure 3. Charging Module with Custom Case & Magnets

Figure 4. Computer Vision for Device & Charging Light Detection

Gantry System

Power

Supply

Camera

Motor

Drivers

Raspberry Pi

Figure 6: Overview of System Design

The software ecosystem �consists of three

interconnected applications:

an iOS app, macOS app, and web interface. When installed, the iOS app generates a unique device identifier and streams real-time data to Firebase, including battery level, charging status, thermal state, and timestamps. Both the macOS and web applications pull this data from Firebase, enabling users to monitor comprehensive charging metrics across all their registered devices.

2 of 2