CaneYouSeeMe

Person Holding Smart Walking Cane

Inspiration

We wanted to focus on improving quality of life for the disabled community while incorporating both hardware and software in our project. Accessibility stood out to us the most among the requirements so we were heavily inspired to create a walking aid for people with impaired vision to augment their navigation.

What it does

Our smart walking cane utilizes a small LiDAR sensor to scan ahead of the ground from the cane's position. We include bluetooth and wifi capabilities to link it to our companion app, which alerts users of obstacles, downward slopes and possible ledges via phone notifications.

How we built it

We utilized Swift for our app; HTML, CSS and Javascript for our webpage. We connected our LiDAR sensor, and our ESP-32 board together using a power bank as our power source. We built all of our hardware on a repurposed tripod, which we used as our walking cane.

Challenges we ran into

Our first challenge was having the right hardware for alerting users, when our vibration motor was too weak. We decided to pivot towards alerting the user via phone instead. We also ran into issues securing the hardware to our cane, and making sure the hardware can reliably communicate with each component. We also struggled with the sensitivity of the LiDAR readings and we have to do multiple live tests to improve and recalibrate the detection.

Accomplishments that we're proud of

The transition from acquiring the raw data from the LiDAR sensor, to collision/edge detection was pivotal for us. Creating a functional walking aid in one day is gratifying to our team as a whole, from the challenges we faced to our solutions, both small and large, we're proud to unveil our project.

What we learned

Combining both hardware and software in a hackathon is difficult but rewarding, especially figuring out how to use the raw sensor data to create a reliable detection system.

What's next for Group 8 - IBM

There are so many features we had to cut due to time, like incorporating a camera that captures user walking paths to detect obstacles and ledges within video streams images via a deep-learning object detection model.