Category: Uncategorized

We consolidated all of our testing results so we can have easy to read statistics at our demo.

Finished our poster for demo day.

Prepared our setup for our demo activity:

• Bike on a trainer
• One video showing our testing videos
• One video showing the POV bike riding experience
• One screen showing our overall product statistics, such as reaction times, battery life, waterproof rating, etc.

We started writing our final report as well.

Put together the final presentation and prepared to present the slides. I also gave the oral presentation for CycleSafe on Monday.

We did more testing this week, especially focusing on fixing our blind spot detection and smoothing algorithms to avoid random flickering for alerts. We rode the bike around many parts of Pittsburgh on real streets, and have good footage to review. We will also have to analyze these videos for testing data in our final report.

We are preparing for the demo, and putting together videos/graphics/demo activities that will be fun for the audience to participate in on Monday.

• Tested blind spot detection multiple times. Had to think about the algorithm and try different things because some of the methods for blind spot detection we were using weren’t performing well.
• Implemented blinking for turn signals and potentially for proximity sensors as well.
• Improved the brake light algorithm to be smoother.
• Took the bicycle and jacket out for an actual use test run.

• Constructed bike with Mike. Worked on the bus connection between the front sensor cluster and the back sensor cluster of the bicycle, including support for additional sensors if necessary and backup wires.
• Wrote better startup code for system which runs a full system test before running, starts up with the Pi, and waits to reconnect to the jacket if unsuccessful.
• Made the communication protocols more robust. We were getting issues where the serial interface with the Bluetooth chip on the Arduino was going out of sync. We solved this by using some magic sequences to ensure every message is properly aligned.
• Got bicycle clamps to mount the constructed system onto the bicycle.
• Got the trainer for demo purposes.
• Tested frontal collisions. Ran tests on the mall in front of Doherty Hall. We marked the walking path with distances, set up recording devices, and ran the frontal collision tests to determine both response time and alert lead-time, as well as determining false-positive tests on receding objects.

Sid Lathar, A1 | Progress Report

Accomplishment:

Communication with the Pi was established in the week of April 14th using RFCOMM sockets.

Communication codes were established that the phone would send to the Pi to adjust settings on the bike.

UI was implemented in the App.

Changes to schedule:

The progress is on schedule.

Upcoming work:

Re-work communication codes to allow for 1 byte communication with the Pi.

Got the demo up and running for Monday, but had to use a lot of tape to put things together quickly and allow for repairs.

I spent the next few days making the bike construction more robust. We finally got our bike clamps, so I added screw holes to the boxes and did a screw + hot glue combo to secure the boxes onto the bike.

I also built the turn signals for the bike. I decided to make a nice circular mount for the turn signals, and engrave a pattern on it to make it prettier and more usable. I also added an indicator LED so you know when your turn signal is on.

Finally, I helped with testing the bike this weekend. I’m going to prepare for the final presentation on Monday.

We got to demo our full product this week.

The first half of our week was dedicated to making our physical construction more robust, so removing the need for tape and clamping our boxes onto our bike.

The second half of our week was spent on testing, where we ran collision tests, blind spot tests, and other related tests for CycleSafe.

We are preparing for our final presentation next week.

• Soldered the protoboards and the components to be used for the construction of the system on the bicycle.  Obtained the necessary equipment and supplies for the final connection. Designed the wire “bus” for communication between the systems on the front and the back of the bicycle. Measured out the various parts of the bicycle with Mike to determine how to arrange everything, designing the mounting system “box” and the necessary ports.

• Improved the collision detection algorithm to make it significantly more stable. Now measures velocity over longer periods, allowing for better robustness and better response to errors.
• Tested the 3 LIDAR system using a single I2C bus, by modifying addressing of individual LIDARs. Made use of the LIDAR enable lines to allow the Pi to restart the LIDARs as necessary.
• Implemented the accelerometer for detecting braking, instead of using the reed switch/timing combination as the response time of that was too slow for brake lights. The reed switch/timing combination remains just to measure speed. The Arduino was necessary to read the analog accelerometer, so it paid off to keep the Arduino in the system even though it was technically not necessary, because of its ability to read analog inputs.
• Modified the communication protocol between the Pi and the speedometer Arduino to make it more robust.
• Worked with Sid to design the app interface and determine what features should be included, especially testing hooks.

I assembled the entire jacket, with waterproofing, cable management, and fully soldering all components. We tested the jacket, and all functionality seems good. The only part we’d add to the jacket is perhaps a case for the battery and board circuit.

Images below from top to bottom: the jacket board, debugging the circuit, heatshrinking for cable management and waterproofing.

I have been working on the rest of the Solidworks models. There are a lot of components, but so far it seems like they are fitting together nicely. We decided to have 3 main boxes on the bike. The main challenge with the models is that we have precise angles for the sensors, which takes more precise modeling. Fortunately, Solidworks provides the tools to fully define everything to precision. Below is a view of the back box of the bike.

I have only fully finished the back, but the front box should have similar components to the back box, so that should be easier to do. The 3rd box contains the battery, which is just a simple box with USB cable cutouts.

I will laser cut all the components this week and assemble everything. Then, I will mount the boxes onto the bike, which should complete the construction phase of the project, leaving us a couple weeks to do testing.