AIM Navigator

Autonomous Image-based Machine learning navigator (AIM) previously known as ICON is a fully autonomous delivery robot that navigates using vision only; no GPS, LiDAR, or external infrastructure.

Project Updates

Oct 6, 2025 — Oct 13, 2025

  • Identified 3 possible approaches for the ICON vehicle body
  • After comparing pros/cons, we decided to move forward with designing our own chassis.
  • Also began developing a materials/parts list.

Oct 13, 2025 — Oct 20, 2025

  • For hardware, we decided to move forward using Raspberry Pi Cameras.
  • Next steps are to design a chassis and finalize the part list.
  • Split project into two aspects; hardware and software.

Oct 20, 2025 — Oct 27, 2025

  • Received Raspberry Pi 5 main controller
  • Remote access and initial configuration complete for Raspberry Pi
  • Began designing out a solid model for the car shape.
  • Researched into stereo depth mapping for obstacle distance.
  • Also began research into collision avoidance by using side cameras.

Oct 27, 2025 — Nov 3, 2025

  • Created initial project structure
  • Built initial decision arbiter logic
  • Created test suite for decision making

Nov 3, 2025 — Nov 10, 2025

  • Completed project website structure
  • Rebranding and name change from ICON to AIM
  • Refactored design to incorporate stereo-vision and multiple cameras
  • Created a prototype solid model with plans to start printing some aspects for fit checks with hardware.

Nov 10, 2025 — Nov 17, 2025

  • Testing of mecanum wheels for planned prototype steering mechanism
  • Successful stereo-vision proof-of-concept achieved
  • Placed order for parts required for assembly

Nov 17, 2025 — Nov 24, 2025

  • Completed decision arbiter logic for prototype demonstration
  • Classification and Intelligent Response (CAIR) classifier implemented
  • Stream 1 for depth and stream 3 for classification successfully merged into decision arbiter logic

Nov 24, 2025 - Dec 1, 2025

  • Project website updated to include all members
  • Completed Solidworks CAD design for car frame

Dec 1, 2025 - Dec 8, 2025

  • 3D-printed prototype for camera mount
  • Presented object classification model using simulated dataset
  • Event-triggered classification lowered CPU usage from 90% -> 18 %

Meet the Team

Arianna Delphino

Electrical Engineering

Korben Widmer

Software Development

Nolan Tuttle

Software Engineering

Mathew Hobson

Software Engineering

Arie Gerard

Software Development