25% of car accidents in the US are caused by texting and driving. Although I think the “real solution” is self driving cars or some other great technology, I wanted to run an experiment where, every time you look down to text and drive, you get shocked.
The video explains some of the technical details and you get to see the hilarious experiment I ran but I'll highlight a few other key points.
- I thought a lot about different ways to detect when I look down. Originally this was just a solution for pausing the TV when I look down to eat so I don’t miss my show (lol), so I thought about using a camera to track my eyes. I started to go down this path but quickly aborted to avoid high costs and an over engineered solution.
- I eventually used the MetaMotionC sensor from Mbientlab which has an accelerator and gyroscope, among other things (more info here: https://mbientlab.com/store/metamotionc/). Their documentation wasn’t great so it took some time to get the sensor working with the Raspberry Pi. The sensor clips onto my glasses, so it was pretty simple to tweak the sensitivity settings and change it from detecting when you eat food to when you look down to text.
- Out of curiosity, does anyone have other use cases for what this glasses sensor could be useful for?
- I thought about using the Pi to send a signal to the collar to trigger the shock. This seemed pretty daunting to reverse engineer. I know I’d need some sort of transmitter to send the signal, but how else might I go about reverse engineering it to send the correct signal to the collar? Any tools, tutorials, etc that people recommend I look into? Might be useful for a future project but for this one, I used a solenoid to press the button on the remote.
- Using a solenoid to press the TV remote/shock collar remote buttons was trivial for me since I recently built a robot that plays Guitar Hero (powered by solenoids). I had struggled and learned a lot in that project so there weren't any surprises this time around.
- The actual texting and driving experiment only took about 30 minutes to run. But making the device, setting up the driving course, filming, editing, etc, made this all take about 6-7 weeks. Similar to my Guitar Hero robot, it probably cost around $1,000 for everything (traffic cones are super expensive plus smaller costs like Raspberry Pi, shock collar, and misc tech parts).
1 comment
[ 4.4 ms ] story [ 12.3 ms ] threadThe video explains some of the technical details and you get to see the hilarious experiment I ran but I'll highlight a few other key points.
- I thought a lot about different ways to detect when I look down. Originally this was just a solution for pausing the TV when I look down to eat so I don’t miss my show (lol), so I thought about using a camera to track my eyes. I started to go down this path but quickly aborted to avoid high costs and an over engineered solution.
- I eventually used the MetaMotionC sensor from Mbientlab which has an accelerator and gyroscope, among other things (more info here: https://mbientlab.com/store/metamotionc/). Their documentation wasn’t great so it took some time to get the sensor working with the Raspberry Pi. The sensor clips onto my glasses, so it was pretty simple to tweak the sensitivity settings and change it from detecting when you eat food to when you look down to text.
- Out of curiosity, does anyone have other use cases for what this glasses sensor could be useful for?
- I thought about using the Pi to send a signal to the collar to trigger the shock. This seemed pretty daunting to reverse engineer. I know I’d need some sort of transmitter to send the signal, but how else might I go about reverse engineering it to send the correct signal to the collar? Any tools, tutorials, etc that people recommend I look into? Might be useful for a future project but for this one, I used a solenoid to press the button on the remote.
- Using a solenoid to press the TV remote/shock collar remote buttons was trivial for me since I recently built a robot that plays Guitar Hero (powered by solenoids). I had struggled and learned a lot in that project so there weren't any surprises this time around.
- The actual texting and driving experiment only took about 30 minutes to run. But making the device, setting up the driving course, filming, editing, etc, made this all take about 6-7 weeks. Similar to my Guitar Hero robot, it probably cost around $1,000 for everything (traffic cones are super expensive plus smaller costs like Raspberry Pi, shock collar, and misc tech parts).