Youtube daily vehicle Sep 3 2018

From startups to tech giants, home appliances to self-driving cars, the internet of things

and automation are the tech industry's next revolution.

With more research and opportunities than ever, the demand for STEM innovators with

imagination continues to grow.

We, the next generation of developers, designers, analysts, and engineers, are ready to make

an impact.

As a group of high school students, we're working to hone our programming skills by

working on one of these ventures: the autonomous car.

Our program focuses on providing a unique industry-like experience with an emphasis

on self-directed learning: we work together to plan, prioritize, research, and execute

stages of development, with assistance from advisors only when sought.

Most students choose to participate in the first year program where they develop a product

of their own choosing, mastering essential methodology and teamwork skills.

This timeline encourages the consistent improvement of knowledge, while also catering to individual

skill levels.

Because of our large team, division of labor is a critical aspect of the development process.

To make efficient use of our limited budget and resources, we only use a single low cost

modified remote controlled car and rely heavily on simulation software for development.

By utilizing an extensive abstraction layer, our team is able to test and make changes

even without access to the car.

This also allows for dissimilar hardware, opening multiple opportunities for other designs

in the future.

Last year, we tested different approaches to pedestrian detection, culminating in a

refined image recognition program that could distinguish moving humans from other objects.

We've expanded that code to identify stop signs and other road indicators.

To do this, parameters filter out shapes that could be a sign, before color blobs are identified

within those frames.

If the object is a match, it triggers braking.

We started this year with a basic approach to steering, leaving more advanced technologies,

such as neural networks, for future years.

Using computer vision, the program finds lanes to the left and right of the frame, and takes

the average of pixels along each horizontal axis to get the midpoints.

Average the midpoints for a steer point, and then calculate the steering angle.

After some performance tweaking and final bug fixes, it was time for our presentation

at Portland State University.

We presented on our process, progress, and then, with our fingers crossed, started our

demo.

Although the road was a little too narrow, the car stayed on course.

Our audience was pleasantly impressed, while we were ecstatic.

Our alumni have gained valuable experience within the program, applicable not only to

future endeavors in technology, but every industry.

Thousands of students don't get the opportunities we've had.

That's why we're seeking to enable motivated students by creating and marketing a low-cost,

customizable self-driving car kit.

With your help, we're hoping to put this kit in the hands of students across the world.

We're excited to be empowering the next generation of doers.

You can too.