Abstract
Driving is the most popular form of commute in the United States. Over the past couple of years, much research and capital has been poured into the concept of vehicle autonomy. Vehicle autonomy occurs when a driver is no longer needed in any driving scenario. Tesla and Waymo are currently the leaders in vehicle autonomy, but neither are close to a fully autonomous vehicle. The STS research project identifies how making the collected metrics of the vehicle more “human-like” could be the key to progressing the technology faster. Specific ideas are introduced and the results of their testing are discussed in the thesis.
The technical project is almost the antithesis of the STS research project. The technical project explores how we can take human error out of a driving experience to make it more consistent. Instead of making a vehicle autonomous, we are instead making an automated system to bring the clutch cable of an engine to a bite point, then drop the cable to initiate power to the wheels. Although automated systems are very important in the route to vehicle autonomy, the idea of automatically dropping the clutch was solved years ago with the creation of the automatic transmission.
The STS research project was motivated by personal observations on the current state of autonomous vehicles. Although companies like Tesla have been working on vehicle autonomy for a while, it seems as if no real progress has been made. For years it seemed like the industry was at a standstill, so research was conducted to figure out what next steps should be to progress the technology. There were many “human-like” metrics that were explored in the project. The first idea explored proposed the idea of including traffic control systems to the “vision” of the Autonomous Vehicle. New traffic control systems with cameras on them, capable of communicating with AVs and extending the effective range AVs cameras. Another idea explored was parsing previous collision data to help the AV identify risky areas in a neighborhood. The AV could then react accordingly and be more cautious in these areas. The other approaches discussed how SAE level 3 autonomy could be improved. The idea of the AV learning certain edge cases from the driver which it could apply to similar cases later on was explored. Another idea of drivers beaming into the car was also researched.
The motivation behind the technical project was to help reduce inconsistency among drivers in UVA’s FSAE club races. There is a large degree of variability of experience among drivers. Our group hypothesized that inexperience with the manual clutch of the cart lead to differences in finishing times in races last year. To help combat this inconsistency, an automated clutch dropping mechanism was designed. This mechanism would bring the clutch cable to the bite point after pressing one switch, then would drop the clutch completely, putting the cart in gear after pressing a second switch. Essentially, we are bringing a 120lb cable to two different points with two different switches. Many mechanisms were proposed, but an approach of a linear actuator to hydraulics was chosen due to its favorable gear ratio and programmability. The resulting system was put on a motorcycle to test it and worked after some troubleshooting. A driver was able to go on the motorcycle and put it in first gear upon initiation of the two switches. We were not able to put the system in the cart, this will likely happen through the club or with a different capstone group next year.
In both the technical project and the STS research project, everything was accomplished that I set out to do. For the technical, we wanted a proof of concept of how a automated clutch system could work on the FSAE cart. In the end, we had a consistent system that was very easy to program. The club can choose to do with our research what they want, since ultimately the cart is theirs to change. I look forward to seeing what they do with it. As for my STS research, I found some solid solutions that will help propel the industry forward. I believe that car manufacturers will be implementing some of the proposed ideas over the next few years. With the current data being collected, an SAE level 5 AV will not be possible for a while. However, by changing the type of data collected, the technology could progress much faster.
