Our Mission

Currently, visually impaired people understand their surroundings through the use of a walking stick, a guide dog, or another active form of input. As such, they are currently limited to active or indirect forms of environmental input. VisionVest aims to create a passive and direct way for visually impaired people to avoid immediate obstacles. By leveraging sensory substitution, a tactile vest with vibration motors is used to give the user feedback on their surrounding to assist them with avoiding obstacles.

Mechanical

Consisting of hand woven 2-ply cotton for maximum comfort. 3-layer system for waterproofing the electrical equipment and damping the external motor sound. Custom casing for housing all the moving equipment.

Hardware

An array of vibration motors will map out the data from the camera sensor, creating a haptic model of the camera’s FOV. Powered by high capacity LiPo batteries and a Raspberry Pi 3B+, the electrical system remains fast, reliable and long-lasting throughout a day of use.

Software

A Time-of-Flight (ToF) sensor is used to create an environmental depth map, covering distances up to 6m. The data is mapped to motor frequencies that allow the user to gain sensory enrichment of the environment.

Progress

Weekly Updates for VisionVest

Week 1 - Regrouping

Jan 11, 2019

After coming from the winter break, the team decided to catch up on the project from where we left off. From there, the team finalized the hardware component as the vibration motors were switched from the external rotating to button style motor. Created a critical failure map that could potentially impact time to completion. This helped in what segment to target first to start building the first prototype. Moving forward to building the first prototype from Week 2, as the parts start shipping in.

Week 2 - Building out Individual Sub Systems

Jan 22, 2019

After regrouping, the team has finally started building the first version of VisionVest. The vibration motors need to be tested by January 23 2019, and if acceptable, need to be ordered in bulk. The team also needs to check for CSA approval. In addition, the PICO data needs to be made visible on the Raspberry Pi. For the mechanical side, a first prototype needs to be created with the motor mounds. With all the parts being individually implemented, the aim is to regroup in coming weeks to integrate and have an end-to-end prototype.

Week 3 - Plan to Integrate

Jan 30, 2019

Last week the team worked on building the first version of the prototype. From the mechanical side, the motor mounting design was implemented in a grid form.The electrical team worked on get a 3X3 motor set up with the shift registers and motor drivers. This 3X3 setup is currently being testing with motor mapping software. The software team worked on retrieving raw camera data and converting it to a depth map. Going forward, the goal for this week is to integrate the 3X3 setup with the depth sensing algorithm to show a proof of concept. The group aims to have the Version 1 ready for the meeting with the advisor next week.

Week 4 to 6 - Rapid Prototyping

Feb 19, 2019

The aim for the two weeks was to develop a prototype that works end-to-end. The idea was to have a 3x3 motor matrix, representative of the vest, that would vibrate according what the pico sensor sees. From the software standpoint, the algorithm compressed the 76x48 data array to a 3x3. A paper piece of paper was used as the vest. The electrical team used breadboards instead of PCBs (which had not yet been designed). After the initial prototype was developed and displayed to our advisor, we appropriate received feedback. The electrical team started to develop the PCB for motor control. The software team changed the compression algorithm to appropriate dimensions (6x8) and leveraged shared memory objects to transfer data between the compression algorithm (master) and the motor mapping algorithm (slave). Additionally, all group members helped with soldering motor & LED boards that would make up the vest. Throughout these weeks, a lot of progress was made, and individual parts that are to consist of the final product were being assembled.

Week 6 to 8 - Building Out the Final Prototype

March 4, 2019

Once the t-shirt prototype was built, there were issues with the usability of the product. How does one wear and/or take off the t-shirt? Would it better to have something that is a zip-up or buckled product? How can you counter the warp of the t-shirt with human movement? Hence, to answer these questions a new idea was proposed, that was to create a foam enclosure vest that is buckled from the sides. Very similar to a “bullet proof vest”, this opened a whole new door as new design challenged occured, such as wire management, motor placement, and noise dissipation due to foam now coming in the product. For the electrical side, the PCB boards were finished being designed and ordered. Within a week the order arrived and they were soon populated by our team. In addition, the electrical team connected all the wiring to the vest. The vest was functional, but seperate parts such as the battery and Raspberry Pi were not mounted. 3D parts were printed for the vest.Additionally, although the vest was functional, there were tasks related to aesthetics that still needed to be worked on. From the software side, code was written to simulate a heatmap of the input data and the compressed data for the symposium.

Week 8 to 10 - Polishing and Software Calibration

March 15, 2019

Although the product was functional due to the hard work done in previous weeks, there was still work left to be done. The mechanical team collected the 3D printed parts and began assembling them on the foam vest. The vest was also spray painted to add to aesthetic appeal. The straps were adjusted and enclosures for all components were made. From the software perspective, noise reduction from the sensor was improved, and calibration was done to ensure that there is no drift in the data. The final meeting with the advisors was conducted with the refactored design, and small improvements to improve robustness of the product were made. In addition, the team worked collectively on the final year symposium poster, and brainstormed ideas for the final presentation demo. The team has also set apart time after the final presentation to have peer evaluation and reflect on the the project, with every member receiving valuable feedback. The group is excited to showcase the hard work we have put into designing and developing VisionVest.

TEAM

Meet the VisionVest Team

MAHARSHI PATEL Mechanical Lead

Maharshi is responsible for building the vest by integrating the hardware and mechanical components. He is also aiding in the final packaging of the vest for the best customer experience.

RAJ LAD PM/Software

Raj is working on the software algorithms for depth sensing and motor mapping. In addition to that, Raj is also charge breaking down the project in the weekly deliverables for everyone in the team to execute on .

SHAHZAIB GILLSoftware Lead

Software lead. Shahzaib is in charge of coming up with computer vision solutions for depth detection. He is also in charge of coming up with algorithms that will be used for to convert the depth data into motor mappings.

MICHAEL RUControls Lead

Michael is working on leading the integration of combining the software element to the hardware components. In charge of designing the bridge component for mapping the motor to Raspberry Pi.

KEVIN CAIElectrical Lead

Kevin in working mainly Electrical design for the PCB that will be used for routing all the moving electrical parts. Kevin is also working on the controls and low level aspect of the motor mapping algorithm.