Published on April 21st, 2019 | by Jennifer Sensiba0
Dear Elon, I Have An Idea For Autopilot
April 21st, 2019 by Jennifer Sensiba
What follows is a letter about an idea I’ve been thinking about for a couple of years that could both reduce the impact of electric vehicles and make them go longer (both in terms of range and longevity).
Dear Elon Musk,
First off, thanks for doing what you do with Tesla and SpaceX. And all of the other efforts. If you’ve read some of my other articles, you know how I feel about all of that.
I wanted to share an idea I’ve been kicking around in my head for a couple of years. It could help reduce electricity usage when running Autopilot, which would obviously reduce the environmental impacts, but could also give much greater range. Less wear on the battery would give the vehicle a longer battery lifespan (especially for fleet AP cars, like for a ridesharing service). It also could help people reach their destinations faster by reducing the number of charging stops needed.
I know Tesla has open patents, so I feel comfortable sharing this idea openly like this. I know it will be used to help the maximum number of people who could be helped by it, and I’d love to see anybody implement these ideas as long as they don’t try to keep others from doing it.
I know your time is in high demand, so I’ll get straight to it before telling the story about how I arrived at this for the other readers. The short version is that I’ve been experimenting with different driving methods for decades, and I’ve tried to boil my driving down to some very simple sets of rules that could be automated.
In short, I propose implementing automated hypermiling. I’d call it Super Range Mode, but that’s not ultimately my call.
Super Range Mode Summary
When a user activates Super Range Mode, it could give them a slider. On the left end of the slider would be the most mild hypermiling, implementing minor things to increase range. On the far right, the power-saving strategies would be extreme. Most users wouldn’t drag the slider all the way right, but for emergencies, it could make the difference between arriving or getting a tow. In really big emergencies, like widespread power outages during disasters, it could save lives by avoiding strandings.
When a user lets go of the slider on a setting, the vehicle’s trip planner could estimate the time each setting would require so the user could pick the best balance between speed and efficiency for their needs. In some situations, there could be a setting that does enough hypermiling to eliminate charging stops more than it slows down by, which could get somebody to their destination faster.
For the mildest Super Range Mode setting, the vehicle just does things to optimize range without making any changes to speed. Air suspension (for vehicles equipped) would lower the car. For dual motor vehicles, the drive unit with higher gearing would get most of the power. Chill mode would be activated. Basically, Range Mode, but a little more extreme.
For the next stop on the slider, the trip planner searches for lower-speed routes the vehicle can take to save power while not deviating from typical speeds. The vehicle would try to take access roads, frontage roads, and rural routes where possible. In cities, the vehicle can choose lower speed routes with lower traffic for better average consumption. Finally, the trip planner can be geared toward arriving at Superchargers at a state of charge that allows the highest charging rates and avoids charging during the taper.
For the remaining slider positions going to the right, the program is relatively simple. Different slider positions correspond to different Wh/mile average consumption targets. The vehicle then achieves the goal consumption by allowing the speed to vary with terrain. When on level ground, the vehicle goes the maximum speed it can at that consumption rate, but only up to 5 MPH over the speed limit, and power is cut back at that point to hold that max speed (and save power).
When a downhill is encountered, the vehicle holds max speed, using regenerative braking only at speed limit + 5. Doing this allows for the vehicle to gather up more kinetic energy from gravity and save it as kinetic energy for the next uphill drive without suffering conversion losses that come with regenerative braking. On uphill portions, the vehicle allows the speed to drop to keep the desired consumption rate, but only to 5 MPH below the speed limit, at which point it allows for more power to not go too slow for safety.
Slider settings further and further to the right allow for lower average speeds, lower speeds on uphills, and higher speeds on downhills (up to speed limit +5). At the far end, the speeds would be very slow to the point where most wouldn’t want to risk it, but would still be available for emergency use with user warnings about going too far below average speeds.
Autopilot is probably smart enough to determine average surrounding vehicle speeds and determine the “flow of traffic” speed on a given road when there are other vehicles. This could be used to set safety limits on how slow or fast the vehicle can go to save power except in the furthest right slider position, which would warn the user that this is for emergency use only. None of this would really matter if there were no other traffic on the road, like one might see at 3 AM on a rural highway, but the AP could probably be trained to pull to the side for passing cars on two-lane roads, for safety.
Other Automated Hypermiling Ideas
Here are a few more ideas I have that are far less specific and would take more work to achieve. I’d love to be part of such an effort, but I’ll leave that up to you. The ideas are still free either way.
The vehicle knows what terrain is coming up based on the Trip Planner data. This data could be used for even more efficiency without sacrificing much time. For example, the vehicle would know that the downhill portion ends soon and goes back uphill, and could “cheat” a bit by allowing more speed than speed limit +5 momentarily to gather more kinetic energy for the coming climb. When going uphill and approaching the crest of the hill, the vehicle could momentarily allow more extreme speed drops in anticipation of the upcoming downhill that would quickly gain the speed back for “free.”
I don’t know how far ahead Autopilot “sees,” but I know from experience that it starts braking a lot later than I would. Hypermilers try to cut power earlier and coast for as long as possible when they know they’re going to have to stop. Autopilot could probably be trained to look ahead and coast more to save energy before applying regenerative braking.
Another thing the vehicle could do in low or no traffic is the “pulse and glide.” The vehicle could “pulse” up more speed at the electric motor’s sweet spot of best specific efficiency, and then cut power and coast for a few. In conjunction with terrain data, this could be highly effective.
Finally, it may be useful to hire some hypermilers to drive Tesla vehicles around and gather data for a fork of the Autopilot program that mimics the humans who are best at this. A neural net trained to mimic the best hypermilers could give that skill to everybody as needed, and probably work better than my Super Range Mode idea once it gets enough training.
One Last Idea: Attachable/Detachable Efficiency Accessories
I know Tesla sells a few accessories for the vehicles, such as the carbon fiber spoiler for the Model 3. I’m particularly impressed with the way there are mounting jig/templates used to install them.
I know that people probably wouldn’t want to use funny looking add-on parts all the time, so suction cup and/or straps could be used to temporarily attach pieces to vehicles to increase highway efficiency to lower drag and decrease power usage. For example, a special spoiler made for efficiency rather than downforce could help. Detachable rear wheel well covers would be another plus. On the most extreme end, a hitch-mounted full boat tail could possibly bring a Tesla down well below 0.20 drag coefficients when installed, and maybe even provide extra cargo space for vacations.
How I Arrived At These Ideas
This section is mostly aimed at making this article more interesting for readers, but if you have the time, I’d be honored to hear that you read it. I’ll try to make it fun and entertaining. 🙂
I grew up around a transmission shop. My dad mostly worked on transmissions for individuals and for fleet businesses. When I was too young for school, my dad converted one of the rooms off the side of the garage into a room for my sister and I to play and watch TV while my mom helped with the business in various ways. I learned a lot about how cars work, what makes them go, and what makes them not work.
Later, my dad got out of the automotive business, but we stayed a family of automotive enthusiasts. My grandfather and my dad also worked for years in car dealerships, so we are more familiar with the ins and outs of the business side of things than most. Even after everybody was doing other work, we still had cars and trucks all over in various stages of projects and, to my mom’s dismay, abandoned projects.
I don’t know why, really, but I got into small cars while the rest of the family was into trucks and SUVs. I don’t hate big vehicles, and I’m perfectly competent at driving them forward or backward, with and without trailers. I just have always liked smaller vehicles for fun. It just made more sense to get the power in a smaller package. Why not?
I was also not terribly happy with the idea that you had to choose between performance and efficiency. It just didn’t ever really compute for me. I had a lot of long talks with him about what makes cars get better mileage, what makes them faster, and what things could do both. I learned a lot about gear ratios, cams, intakes, exhaust, and many other things. We even talked about exotic engine designs and alternative fuels quite a bit. I remember once, when talking about electric vehicles, my dad said, “People think electric cars are slow, but I can tell you that there are electric motors that would snap your neck. The problem is batteries. You might be able to get from here to the store, maybe back if you’re lucky.”
At one point, I decided to build a late 70s Chevrolet Monza with a small block chevy and a 4 speed transmission with torque converter lockup and overdrive. The idea was to have a small, lighter vehicle with plenty of power, but optimized to get terrific highway mileage.
But then, something caught my eye that would drive my dad nuts for years. One day, as a younger not-driving-yet teen, a really cool car passed us on the interstate. It was low to the ground, had popup headlights, and it was pretty obvious from the shape of it that the engine was behind the two-seat cabin. It was sleek, had a pretty cool spoiler, and said “PONTIAC” really big on the back tail lights, lit up like a neon sign. The econo-nerd in me and the “I like cool cars” part of me both reacted to it in the same way. I wanted one. My dad was able to tell me all about the vehicle. It had the same transmissions as front wheel drive GM cars, and was available with either the 2.5L “Iron Duke” engine or the 2.8L 60-degree V6.
Once I ended up getting one, I started all sorts of little experiments. I knew that mine had the 3-speed auto without overdrive, and that it would get better mileage at lower highway speeds. I experimented with different routes to high school (35 miles each way), different speeds, and different strategies. I had a custom exhaust built with far larger pipes and fewer restrictions. I changed the intake to pull outside air in much more directly. I even fixed up the EGR system, and saw that it gave better mileage. I tracked every tank religiously. I was typically getting 30 MPG, for a 1986 car.
Later, I had to switch vehicles to have room for family, but was pretty surprised how far things had gone from 1986 to 2003. The Cavalier I had was about as powerful as the Fiero I had before, but got almost 40 MPG on the highway. I ended up doing a lot more learning about more modern engine technology.
Later, after college, I got another Fiero. This time, I was smart and got the 5 speed with overdrive. I had some money, so ended up doing a lot to it. Better exhaust, better intake, and I managed to get some manual control over the EGR system to get that car to give me almost 40 MPG. Life eventually took me in other directions, but I still kept that interest in clean auto technology.
More recently, I realized that the things I did as a teen and just after college were something that other nerds did. I learned about ecomodders, hypermilers, and other such things. With my past knowledge, I stumbled my way into the emerging world of electric vehicles. The only thing I could afford, though was a 2011 Nissan LEAF with a mostly trashed battery. Then, the challenge became range instead of MPG. Living near El Paso, the charging stations are few and far between. I did manage to do a lot of rideshare work with that car, and even would take it on some longer trips that it wasn’t supposed to be able to go on.
I’ve since switched cars a few times and now drive a 2018 LEAF, which I’ve pushed far beyond its rated range on a number of occasions manually doing the sort of driving I propose we now automate. The challenge continues, and I’m still learning new things every day.
I know I’m rambling on and on here, but the big point is that I’m finally starting to find what I was looking for as a kid. I’m hoping that I can keep on experimenting and pushing the limits like I always have. This is just one way to do it.