Dr. Franklin De Santa1, and Trevor Phillips2
1 Department of Street Technology, Cranberry-Lemon University, Pittsburgh, PA, USA
2 Leader of the Master Thieves Heist Guild
Abstract
The car theft business has been an integral part of the American automotive industry for decades. If it weren’t for widespread grand theft auto and stereo theft, we’d all still be driving reliable cars such as the 90’s Toyota Camrys or 00’s Honda CR-Vs. While many fear that the advances in cyber and physical security in newer vehicles may be the end of the theft driven auto industry of inner cities across America, new techniques have been shown to make state of the art self-driving cars even easier to hijack, capture and sell for a profit than their outdated counterparts. The new technology has even driven higher demand for the products from modern chop shops than ever before. This paper will outline methods to confuse, disable and take advantage of the state-of-the-art algorithms and sensors that modern autonomous vehicles are now equipped with to revitalize the domestic car-jacking economy. In this paper, we will test new autonomous vehicle hijacking techniques such as basic traps, artificial pedestrians, laser pointers, and a pheromone induced honey-pot lure.
Keywords: Self-Driving Cars, Cheetah Based Driving Model, Image Classification, Large Galvanized Steel Traps, Cheetah Pheromone-soaked Vespa’s, Autonomous Systems, Grand Theft Auto, Machine-Learning, Animal Behavior, Dead Fall Traps, Chained Antelopes, Free Parking Bait, Large Boulders on a Hair Pin Trigger, Chop. Shop Demand Boom
1. Introduction
It’s a well-known fact that autonomous electric vehicles, especially Elon Musk’s Tesla models, worship coolness over safety so much that they implemented an animal hybrid Cheetah Brain for speed and agility into their autonomous vehicle logic. Little did they know that it would be Tesla’s downfall. With a brain like an animal, the car can be trapped like an animal. Even the safety sensors provide vulnerabilities.
When the green movement swept across the United States, demand for rare earth minerals, such as copper used in electric motors or lithium and nickel used heavily in battery production, has skyrocketed. With most of these rare earth materials existing in the mines of poor South American and African countries that are easily overthrown in coups orchestrated by covert special operation, it would only be ethical to do whatever we can to supply these materials domestically.
Modern vehicles are heavy laden with these rare earth materials and the average chop-shop profit from car models newer than 2016 has been doubling yearly. While physical security and cyber security has been exponentially increasing over the last decade, more physical capture techniques have become widely used since these newer self-driving cars have created wide-open vulnerabilities. This paper will discuss methods to take advantage of two vulnerable systems; safety sensors, and the biology-based self-driving algorithms.
1.1 Safety sensors
Many modern cars now contain safety features that have made driving incredibly safer for their passengers and pedestrians. Unfortunately for the car, that safety has come with a cost. They are very easy to fool.
If it can be sensed, these self-driving cars can see it. However, these self-driving cars do not always know the best way to interpret what they are seeing. For instance, you can make something look like a nice parking spot or a pedestrian surprisingly easily, therefore confusing the car to perhaps act irrationally or exactly how you would want it to behave. Though algorithms have been developed for an autonomous vehicle morality, the results suggested that no moral-safety framework will ever beat the “If (about to hit someone): Don’t;” (IATHSD) algorithm. All the sensor needs to know is that the object, be it a car or a pedestrian on one of those zippy electric scooters people are leaving around on sidewalks, should not be hit.
1.2 Biology-based algorithms
It’s a well-known company secret that nearly every commercially viable self-driving algorithm is based on animals[1]. Using any normal Animal Algorithm Extractor™ such as shown in the following figure, the thoughts of any animal can be digitized and applied to a vehicle’s central processing operating system. The animal characteristics are then utilized by the entire vehicle for speed, agility, ferocity, and wisdom.
Choosing what animal to use for what vehicle can be the toughest design choice you can make when designing a self-driving algorithm[2]. For the techniques outlined in this paper, it is essential to know which animal’s brain has been mimicked to optimize the methods.
Originally, the idea of using a Cheetah’s brain to power the core algorithm of the Tesla was controversial. Many thought it would be too dangerous to incorporate the brain of a predator. But the Tesla developers are well-known for doing things because they are cool, and for some reason they believe Cheetah hybrid electric vehicles are cooler than improved public transportation or better city planning. Luckily, safety algorithms transposed from Isaac Asimov’s three laws of robotics etched into the machine language and the IATHSD algorithm has allowed these predator minds to pass safety tests with flying colors. It was a risk that allowed the consumer to drive the fastest cars on the road. Adapting the Cheetah brain was seamless, the only hardware change necessary was the inclusion of little, tiny whiskers you can’t see or feel on the front of the Tesla to balance the cars body, act as a sort of radar, and communicate emotions to other Teslas[3].
It’s not safe to assume every Tesla is a Cheetah and that every Kia is a Pangolin. For instance, the Tesla Cyber truck is based on the brain of an ant to multiply raw towing power and carrying capacity. While a chained antelope may be a good bate to trap a Tesla sedan, a Cyber Truck may be lured with simple sugar water or leaving open a jar of syrup. The methods for different animals are similar in structure and strategy but the differences in implementation and bating are paramount.
2. Methodologies
In the past, hijacking cars normally involved picking a lock, hot wiring and just driving away. With the introduction of modern car alarms and trunk monkeys[4], not only is this technique difficult but it can be dangerous. It is important to adapt with the time and use more modern techniques for such modern security methods. Many of the newer techniques, while it can take some planning, more materials, and a larger crew, are not only more likely to succeed, but are less dangerous and less likely to be incriminating. Many of the methods are even passive and perfect for the thief on the go. After initial setup, all you have to do is wait and periodically check the traps. With game cameras, the traps may be monitored digitally.
2.1 Basic trapping
When a self-driving car has dropped off its passenger and begins looking for a parking space or while its passenger is distracted with a freemium game, there are many effective means to trap the car in place. It’s a very simple process.
Set up a mechanism with some hair pin trigger which will either trap or incapacitate the vehicle. Bait it with some 10W-40 oil or a nice pine-scented air freshener or even just a free parking sign and let the trap do the work for itself. Figures 2–4 show the most commonly used traps for self-driving cars including a dead drop, bear trap for trapping the vehicle in place and the baited cage for the more humane trappers.
In the following figure, a diagram can be seen in which a large boulder is precariously balanced by a hair pin trigger constructed with sticks and twine:
Nestled under the boulder, you may find a free parking sign. As a human operator would never trust such a sign without a weekly schedule in a crowded city, an autonomous vehicle will have not been traumatized by parking tickets. As a vehicle jostles the sign and trap, gravity powers the boulder in the downwards direction onto the hood of the vehicle, trapping and incapacitating itself to be captured without any resistance. It’s important to choose a large enough boulder to make a clean kill or the vehicle could just be maimed and suffer needlessly.
If you’re okay with the vehicle suffering, you can use a standard five-foot-wide bear trap as shown in the following figure:
By attaching the trap to a concrete parking barrier with a reinforced chain, even if the autonomous vehicle has enough traction control and four-wheel drive to get away, it will remain trapped. Although electric vehicles don’t need oil like gas-powered vehicles, old firmware deep in the car’s motivation circuits don’t know the difference and still crave the stuff!
Finally, a galvanized steel cage may also be baited using pine scent. This is considered the most humane trap for autonomous vehicles preserving the maximum resale value. If they are trainable, they can be repurposed for another owner or the circus. As shown in the following figure, a pressure plate attached to a trap door can trap an unsuspecting electric vehicle searching for that new car smell for its owner. We believe the smell reminds them of their youth in the sales lot, creating a strong emotional bond with the smell.
It’s important to set many traps for the most success. No one trap is a guarantee, but enough traps around a business park is. Methods for building and maintaining these traps are detailed in our previous paper on vehicle trapology[5]. The key is to make the traps look inviting for a car. Try and get it in the shade away from confusing parking signs, shady parts of town, and potholes. Try for a nice wide space where it’s unlikely to be scratched by another car. It is possible to bait traps with pheromones but then you only expose your trap to one particular type of self-driving algorithm and core animal gender. With too many pheromones, the vehicles’ aggressive behavior may be triggered and will damage your traps.
Some traps are clearly more destructive than others. Depending on what you are trying to harvest from the vehicle, you may want to use more humane trapping. However, sometimes using the less destructive methods can intuitively become the less humane trap. There have been several instances reported by Yukon Jack and Jim Baker[6] where “…some such darn vehicles hit the bear trap with their non powered axel and will fight for hours until it frees itself when the axel itself comes off the car and that slippery rascal just drove back to its owner in a panic on its two front wheels.”
Another instance in Yukon Jack’s trapping guide[6] saw a 2018 Prius hybrid ram right through its cage. “Even if it couldn’t escape, those poor fellas take a lot of damage ramming into the steel reinforced cage. I’ve lost many a traps to turbo mode. The thing about these electrics and electric hybrids is they don’t need a lot of space to build torque. In the early days of car trapping, these cars would burn out a gearbox before getting away.” While the dead drop looks the least humane, it’s clean and quick with the autonomous vehicles only suffering for a moment.
2.2 Artificial pedestrians
The second slightly more labor-intensive method for capturing a vehicle is creating artificial pedestrians and swarming. Thanks to the IATHSD algorithm, the safety sensors of a self-driving car will never harm a pedestrian. Not even if that pedestrian only looked like a mannequin dressed in clothes from Good Will; in fact, not even if it was just the general size and heat signature of a pedestrian.
In the early days of trapping safety-enhanced vehicles with pedestrian lookalikes, all you needed were a few cardboard cutouts of pedestrians from drivers ed. courses to strategically place around the vehicle, trapping it in place. Now the sensors are too smart for that. In the vehicle security arms race, the newer self-driving models were equipped with IR cameras as well to determine if something looked like a human and also was as warm as one. Some clever thieves took to heating a large sack of potatoes and duct-taping the products to the card-board cutouts. But, once the potatoes cooled over an hour, most autonomous vehicles adjusted their learning algorithms to prevent such false positives due to the potato’s distinct gamma signature.
Several street fires later, our team gave up on heating the cardboard cutouts. Instead, we found that while it’s easy for the car sensors to detect movement and shape, they are proficient at classifying images like they have been advertised to. This is when we started using hogs and trained chimps to surround the vehicle until incapacitated. This does require much more planning, hog feed, or chimps brave enough to stand in front of a car without moving when the car self-honks. The hog feed to keep the animals in place is something you can easily pick up at any farming supply outlet, but training your animals to not move when honked at is another task. A Pavlovian method developed by Dr. Caesar Millan[7] has been found to be a proven method to keep your animals in place. The method involves a simple reward system while increasing the sound of a horn until they do it on the loudest setting without food. That’s when you’re ready to trap a car with your farm animals.
2.3 Laser pointers
Moving on to active techniques, another effective method of hijacking a self-driving car is the laser pointer. However effective, it is only useful for the Cheetah-based vehicles and autonomy cores based on the animal subfamily Felinae. This is proven in research by ‘Doc’ Antle[8] which shows that the Felinae subfamily or cat-like animals cannot resist laser pointers. Many theorize that when this species developed their bony hyoid allowing purring at the expense of the infamous roar, they also developed a behavior to entertain their owners for food by confusing bright spots of light with prey and entertaining their prehistoric Egyptian patrons who developed the reflective bronze to create such a beam of light.
To execute this method, you will need a bright green laser pointer (the type used at the planetarium) and an interception crew or trap. Once the Cheetah brained Tesla’s see that green spot on the ground, all you have to do is keep the dot in front of the vehicle until you have the Tesla right where you want it.
It is important not to let the dot disappear for very long or the Tesla will get bored and continue on its predetermined route. It’s also important that the dot remains in front of the vehicle. If the dot appears on the hood of the car or directly shined into the sensor, the dashboard sensors will realize that the green dot is not prey and confuse itself until it becomes disinterested. This technique is best used with other passive techniques to guide the vehicle into a trap location. In an open parking lot, you can keep a self-driving car doing donuts in place for fifteen minutes before it gets dizzy and gives up, but it would be too dangerous to approach. That’s when your crew has to be ready with the follow-up trap.
2.4 Honey-pot technique
Similar to laser pointers, another active method of leading a self-driving car wherever you want is the honey-pot technique. Using a hard-to-catch vehicle, you can attach a porous box of pheromones which will attract nearby self-driving cars of the corresponding species and opposite gender. Sometimes, it can help pick a sexy vehicle with a flashy color such as a brightly colored vespa or a VW bug. Sometimes this can be the better method of hijacking your self-driving cars in city streets where you don’t have good open areas to use a laser pointer for long stretches of road. A diagram of the honey-pot technique can be seen in the following figure:
It may be prudent to drive the vehicle remotely for this method. Directions to rig a Vespa or VW bug for this trap technique can be seen in our paper on the honey-pot technique development[9] where the tactic was first developed during the height of car mating season. Remotely controlled vehicles can prevent serious injury in the eventuality that the self-driving car does catch up with the honey pot. You do not want to be in between your inviting car and one ton of lithium battery powered horny Tesla. For those voyeuristic thrill seekers, it is advised to at least wear a helmet and attach a sturdy roll cage. The added weight of a roll cage will, however, make your vehicle less appealing to some self-driving vehicles and you may need a more potent pheromone concoction or some suggestive bumper stickers.
3. Data collection
In order to test these methods, we evaluated passive trapping methods (bear traps, dead falls, steel cages, and pig-based artificial pedestrians) by themselves and then tested combinations of the active (laser pointer and honey pot) and passive methods.
The laser pointer and the honey-pot techniques were tested leading a vehicle into the three different types of basic traps as well as an intersection prepped with farm animals dressed as pedestrians for the artificial pedestrian technique. Traditional traps such as the galvanized steel cage, a dead fall, or bear trap were tested using a variety of different baits.
For the first week of testing, we were getting repeatable results and then observed a sharp decline in trap yield. It turned out that we were over trapping in Pittsburg. We had to move operations to Cleveland where we annotated our data set and began using a catch-and-release method of testing.
4. Results
Results of the new car Hijacking techniques were mixed across the board. It is hard to assess some of the pheromone techniques as another decline in performance was observed as owners began chemically castrating their vehicles after a local recall was issued. It’s additionally difficult to determine the vehicles gender and choose the correct pheromone without looking under the vehicle chassis. Some owners refused the recall out of guilt as their friends’ cars apparently just didn’t drive the same after the procedure and it wasn’t natural. Regardless of your opinion of chemically castrating cars’ autonomy cores, it was not friendly towards our data analysis.
4.1 Trap bait results
The results are shown in mean cars trapped per week over a three-month study. Each trap was constructed at normally distributed locations around Cleveland office parks and other business locations and lunch spots. Traps were checked twice a day and baits were refreshed even if there were no cars in the trap. Results are shown in the following table:
| Dead Drop | Bear Trap | Cage | |
| 10w40 Oil | 1.3 | 1.1 | 0.6 |
| Female Cheetah Pheromones | 3.6 | 3.2 | 1.3 |
| Pine Scent | 0.8 | 0.4 | 0.0 |
| Free Parking | 2.3 | 2.5 | 1.0 |
| Chained Antelope | 10.0 | 10.0 | 10.0 |
With extremely publishable p values (p<0.01), we have analyzed that the Dead Drop is the most successful trap with the bear trap as a close second. Even with the recall, the female pheromone worked third best but was not close to being as successful as the chained antelope. We only checked our traps on weekdays because of Cleveland union pressure in the thieves guild and the chained antelope never failed to trap a car. There were some instances where vehicles escaped their traps, but the antelope kept the vehicles content until we came around to check for the day. Though the vehicle Cheetah brain obviously lacks a digestive track, once an antelope has been consumed via the front trunk of frunk, an electronically adapted hormone is released to induce a satisfying cat nap to digest the fresh antelope meat. With such positive results in between the female pheromones and chained antelope baits, it appears it is true that male Cheetahs really only have two emotions.
In an unexpected result, the bear trap turned out to have regular issues in damaging the vehicle even more than the dead drop. It worked well at trapping the vehicle but it also caused an unmitigated chain reaction. It was observed that piercing the underside of a Lithium Ion battery in the humid Great Lakes area, or worse, above a common street puddle can cause a fairly destructive explosion. Most of the bear traps were triggered by wheels but about 10-15% jumped over too fast and the car would basically explode from the puncture.
4.2 Active method results
The combination test showed much more variety in effectiveness. The results from our tests can be seen in the following table as the percentage of successful traps given each attempt. Overall, the percent rate exceeded all passive baiting techniques except for the chained Antelope.
| Dead Drop | Bear Trap | Cage | Farm Animal Pedestrians | |
| Laser Pointer | 74% | 78% | 44% | 95% |
| Honey Pot | 65% | 63% | 85% | 100% |
In this test, it appeared that the laser pointer was much better at leading the self-driving cars into the dead drop and bear trap while it was difficult to lead an autonomous vehicle into those traps without damaging the honey pot. The honey pot was, however, extremely effective with the cage trap as we could honk and flirt until the vehicle entered the cage from temptation.
The artificial pedestrian technique appeared to be the most effective and only failed once out of our 40 trials when protestors of the Columbus Day Parade spooked all of the animals away before we could even take off the hubcaps. Nearly every autonomous vehicle we tested could not tell the difference between a chicken/pig and a pedestrian, or a goat and a delinquent teen zooming about on a scooter. As soon as we had the vehicle in position, the animals swarmed on the feed, and it was all over. Each model of each car was consistently trapped by their programmed moral compass.
5. Conclusion
It turns out, making cars really fast with the algorithm of a Cheetah brain does have some negative consequences for their owners and not just for the developers who have to debug the thing. The world of autonomous vehicles and grand theft auto will forever advance as technology leaders find ways of making cars faster, safer and more secure. It wouldn’t surprise me if for the next study we have to start using better street parking signs, larger pedestrian equivalents, and more enticing honey-pot vehicles when better advances in the autonomous vehicle logic see through the traps. The field of grand theft auto will never be finished, for it is a cat-and-mouse game one day and a cheetah-and-antelope game the next.
6. Future work
For our next iteration of this street technology research department. We are developing a method to organize Cyber Trucks equipped with ant brain autonomy cores to construct a large mound on the grounds of the Cranberry Lemon Football pitch while only using a cybernetic Ant Queen equipped with a long-range pheromone dispenser. Once enough Cyber Trucks are on the road, we will begin the field test..
7. Conflict of interest
My coworkers all have Tesla’s and they are getting really annoying about how much torque they have and being part of the climate solution.
8. Acknowledgements
Thanks to Dirty Mike and the boys at the car thieves’ guild on sixth street for sharing their advice and experience with my team of scientists during this study. Thanks as well to the owner of the chop shop who did not want to give me his full name but works out of the Goodyear next to the subway on Washington St after hours.
References
- US Patent #201984928, Animal Hybrid Autonomous Vehicle Speed Boost Design
- Cheetara F and Snarf H 2015 Animalistic Design Decisions in Autonomous Vehicles :: Journal of Animal Machine Humunculi
- Maven Bartholomeow, Why Do Teslas Have Whiskers: A Cat-Hybrid Approach to not getting stuck in a Parking Garage :: Journal of Cybernetic Animal Sensing
- Trunk Monkey Bridge Commercial Suburban Auto Group :: Playlist of garbage YouTube videos :: https://www.youtube.com/watch?v=hq0mUxRKHQY
- Dr. Franklin De Santa and Trevor Phillips 2019 Methods in Urban Vehicular Trapology for the Thief on the Go :: Journal of Criminal Futurology
- Yukon Jack and Jim Baker 2018 Humane Techniques for Capturing Rogue Autonomous Vehicles :: Wilderness Journal for the Greater Long Beach Area
- Dr. C. Millan 2016 Teaching Farm Animals to be Dominant not Aggressive :: Animal Whispererer Annals
- ‘Doc’ Antle B. 2014 Applying Big Cat Techniques to Midsize Autonomous Vehicles :: Journal of Big Cat Sex Cults
- Dr. Franklin De Santa and Trevor Phillips 2019 The Honey Pot Technique and How to Seduce a Tesla Model S :: Annals of Street Technology
If you enjoyed this article applying animal trapping techniques to steal self driving cars please like, share, and subscribe with your email, our twitter handle (@JABDE6), our facebook group here, or the Journal of Immaterial Science Subreddit for weekly content.
If you REEEEALY want to read more papers like this, for the equivalent price of a Chipotle Burrito, chips and Queso, you could buy our new book Et Al with over 20 hand picked Jabde articles for your reading pleasure, it’s the perfect Christmas/Birthday gift for confusing your unsuspecting family members! Order on amazon here: https://packt.link/at4bw Please rate and review so that you can brag to your friends about having opinions or showcase your excellent taste in reading material!
One thought on “Novel Techniques for Hijacking Self-Driving Cars”