This week was a productive week for our team. We have no design or schedule changes. However, there are a couple risks that we have to manage. This week we requested access to a very comprehensive object detection dataset Imagenet. However, the website is a little out of data and we are unsure if access will be granted in time. Our backup Dataset is the COCO Dataset. This upcoming week, we will also be working with the Ultrasonic sensors that are very fragile and can easily stop functioning as needed, We have ordered extra sensors in order to avoid any loss of progress.
A was written by Ishan, B was written by Oi and C was written by Ryan.
Part A:
Our product solution will meet a need for public safety. Our system will enable visually impaired people to navigate around any impeding obstacles or objects that they could encounter in an outdoor setting. Our design will filter objects that are not an immediate danger to the user like objects moving away or objects that are not directly impeding the user’s direction of motion. This means that we will prioritize objects moving toward the user and any obstacles that are directly impeding a user. We will do this through the use of ultrasonic sensors that will detect the objects and assert a respective distance from the user. Then, our image detection model will tell the user what object this is so that they have a better idea of the obstacle they’re encountering. So, to surmise the device will tell the user where and what the object is with respect to the user. This device is meant to accompany already existing navigation devices like the walking stick as most visually impaired people are comfortable using devices like this to detect any low-level objects. Our design will help to supplement devices like the walking stick by protecting against high-level/moving objects as well.
Part B:
Our headset is taking in the needs of the visually impaired community through interviews and user testing. We want to create a product that will be responsive to their needs and help them interact and navigate through their environment more easily by helping the users detect and identify obstacles around them easily. We plan to make our product more adaptable to visually impaired users with different needs through the different calibration settings for the headset for different comfort levels. We hope that our tool will be more accessible and also promote inclusivity.
Part C:
Our product is designed so that it isaffordable for most users. This requires us to minimize cost of the materials and the complexities of the build. There we are using low cost materials such as Raspberry Pis, small ultrasonic sensors, a jetson, and a custom IOS app. The app should be free for users. In addition, through the use of a Jetson to run our object detection model, we have minimized recurring cloud expenses (ie. aws). The different componenst we have chosen also inteact with each other at ease, allowing us to simplify the product build lowering production costs.
