Shower Thoughts in Your Car

The subreddit for Shower Thoughts offers wisdom ranging from the profound to the mundane. For example: “Every time you cut a corner you make two more.” Apparently, [Harin] has a bit of an addiction to the subreddit. He’s been sniffing the CAN bus on his 2012 Hyundai Genesis and decided to display the top Shower Thought on his radio screen.

To manage the feat he used both a Raspberry Pi and an Arduino. Both devices had a MCP2515 to interface with two different CAN busses (one for the LCD display and the other for control messages which carries a lot of traffic.

The code is available on GitHub. There’s still work to do to make the message scroll, for example. [Harin] has other posts about sniffing the bus, like this one.

We’ve covered CAN bus quite a bit, including some non-automotive uses. We’ve even seen the CAN bus for model railroading.

Self-Driving Cars Get Tiny

There’s a car race going on right now, but it’s not on any sort of race track. There’s a number of companies vying to get their prototype on the road first. [Anurag] has already completed the task, however, except his car and road are functional models.

While his car isn’t quite as involved as the Google self driving car, and it doesn’t have to deal with pedestrians and other active obstacles, it does use a computer and various sensors to make decisions about how to drive. A Raspberry Pi 2 takes the wheel in this build, taking input from a Pi camera and an ultrasonic distance sensor. The Pi communicates to another computer over WiFi, where a neural network operates to make decisions about how to drive the car. It also makes decisions based on a database of pictures of the track, so it has a point of reference to go by.

The video of the car in action is worth a look. It’s not perfect, but it’s quite an accomplishment for this type of project. The possibility that self-driving car models could drive around model sets like model railroad hobbyists create is intriguing. Of course, this isn’t [Anurag]’s first lap around the block. He’s already been featured for building a car that can drive based on hand gestures. We’re looking forward to when he can collide with model busses.

Continue reading “Self-Driving Cars Get Tiny”

Simple Fuel Pressure Alarm Averts Diesel Disaster

If you could spend a couple of bucks on a simple project that might prevent a $2000 repair bill on your vehicle, you’d probably build it, right? That’s the idea behind this simple low-pressure alarm for a diesel fuel system, and it’s so simple it makes you wonder why the OEM didn’t do it.

We normally see [Bob Johnson] coming up with nifty projects (like this claw or this camera slider) that more often than not combine woodworking and electronics. But no tree carcasses were harmed in the making of this project. [Bob]’s goal is just to sound a warning and flash a light if the output of a pressure switch goes to ground. That indicates the lift pump in his Dodge Ram’s fuel tank has failed, which could lead to the sudden failure of the downstream injector pump for lack of lubrication by the fuel itself. His simple ATtiny85 circuit lives on a small perfboard in a 3D printed case and taps into a $30 fuel pressure switch. The microcontroller code enables a short delay to prevent nuisance alarms, and if the pressure drops below 5 PSI, [Bob] gets a chance to shut down the engine and disappoint his mechanic to the tune of $2000.

Maybe it’s planned obsolescence on the OEM’s part, or maybe it’s not. But kudos to [Bob] for a simple hack that averts a potentially expensive problem.

Continue reading “Simple Fuel Pressure Alarm Averts Diesel Disaster”

Volkswagen Beetle – The Most Hackable Car

If you build a better mousetrap, the world will beat a path to your door. Of course it helps if your mousetrap is reliable, simple, cheap, and easy to work on. In the car world, look no further than arguably the most successful, and most hackable, car in history: the Volkswagen Type 1, more commonly known as the Beetle. The ways in which this car was modified to suit the needs of a wide range of people over its 65-year-long production run proves that great design, ease of use, and simplicity are the keys to success, regardless of the project or product.

Built by Ferdinand Porsche in 1930’s Germany, the Beetle was designed to be a car for anyone and everyone. Its leader at the time wanted a true “people’s car” (i.e. “Volkswagen”) that was affordable for a German family, could reliably travel at sustained highway speeds on the new German autobahns, and easily be repaired by its owners. The car features an air-cooled engine for simplicity and cost savings: no radiator, water pump, or coolant, plus reduced overall complexity. The engine can be easily removed by disconnecting the fuel line, the throttle cable, and the four bolts that hold it to the transaxle. The entire body is held on to the chassis by eighteen bolts and is also easy to remove by today’s standards. There’s no air conditioning, no power steering, and a rudimentary heater of sorts for the passenger cabin that blows more hot air depending on how fast the engine is running. But possibly the best example of its simplicity is the fact that the windshield washer mechanism is pressurised with air from the over-inflated spare tire, eliminating the need to install another piece of equipment in the car.

It’s not too big of a leap to realize how easily hackable this car is. Even Volkswagen realized this and used the platform to build a number of other vehicles: the Type 2 (otherwise known as the bus, van, hippie van, Kombi, etc.) the eclectic Karmann Ghia, and the Types 3 and 4. Parts of the Type 1 were used to build the Volkswagen 181, commonly referred to as “the Thing”. Ferdinand Porsche also used design elements and other parts of the Type 1 to build the first Porsche, essentially making a souped-up Beetle. The rear-engine, rear-wheel drive layout of modern Porsches is a relic of this distant Beetle cousin. But the real magic is what people started doing to the Beetles in their backyards in the ’60s and 70s: turning them into buggies, off road machines, race cars, and hot rods that are still used today.

At some point around this time, a few people realized that the Beetle was uniquely suited to off-road racing. The type of suspension combined with the rear-engine, rear-wheel-drive layout meant that even without four-wheel drive, this car could excel in desert racing. There are still classes in this race for stock Beetles and modified Beetles called Baja Bugs.

Continue reading “Volkswagen Beetle – The Most Hackable Car”

Are Powdered Metal Fuels Just a Flash in the Pan?

It’s no secret that fossil fuels are quickly becoming extinct. As technology charges ever forward, they are disappearing faster and faster. Many of our current dependencies on fossil fuels are associated with high-energy applications like transportation. Since it’s unlikely that global transportation will ever be in decline for any reason other than fuel shortage itself, it’s imperative that we find something that can replicate the high energy density of fossil fuels. Either that, or go back to the drawing board and change the entire scope of global transportation.

Energy, especially solar and wind, cannot be created all over the world. Traditionally, energy is created in situ and shipped to other places that need it. The proposed solutions for zero-carbon energy carriers—batteries and hydrogen—all have their weaknesses. Batteries are a fairly safe option, but their energy density is pretty poor. Hydrogen’s energy density is higher, but its flammability makes it dangerously volatile to store and transport.

Recently, a group of researchers at McGill University in Canada released a paper exploring the use of metal powders as our zero-carbon fuel of the future. Although metal powders could potentially be used as primary energy sources, the transitory solution they propose is to use them as secondary sources powered by wind and solar primaries.

Continue reading “Are Powdered Metal Fuels Just a Flash in the Pan?”

Good News! It’s The Dacia 1310!

Although we’ve never had the privilege to drive one, [skaarj] tells us Dacia made some terrible cars. The Dacia 1310, a communist clone of the Renault 12, was cheap, had sixty-two horses under the hood, and was easy to maintain. The cabin, by all accounts, is a bit lacking, giving [skaarj] the opportunity to improve the instrument cluster and dash. He’s not throwing a stereo in and calling it a day – [skaarj] is upgrading his Dacia with retro-futuristic components including a vacuum tube amp, a CRT computer display, and an unspeakably small dumb terminal.

[skaarj]’s build began with a hit and run accident. With most of the body panels on the passenger side of the car removed, [Skaarj] ground some rust, rattle canned some rust proof paint, and bondoed the most offensive corrosion. Work then began on the upgraded dash, with a few choice components chosen including an old Soviet television, a hardware neural network to determine hardware faults, and a bizarre implementation of a CAN bus on a car without any of the requisite electronics.

This is one of those projects that can go on forever; there’s a lot you can do with the dashboard of a car if you’re not constrained by a suffocating desire to appear normal. In that respect, [skaarj] has this one locked up – he’s got a vacuum tube amplifier and enough CRTs in this car to add retro satellite navigation. It’s a great entry for The Hackaday Prize, as something cool is sure to come out of this project.

The HackadayPrize2016 is Sponsored by:

The Predictability Problem with Self-Driving Cars

A law professor and an engineering professor walk into a bar. What comes out is a nuanced article on a downside of autonomous cars, and how to deal with it. The short version of their paper: self-driving cars need to be more predictable to humans in order to coexist.

We share living space with a lot of machines. A good number of them are mobile and dangerous but under complete human control: the car, for instance. When we want to know what another car at an intersection is going to do, we think about the driver of the car, and maybe even make eye contact to see that they see us. We then think about what we’d do in their place, and the traffic situation gets negotiated accordingly.

When its self-driving car got into an accident in February, Google replied that “our test driver believed the bus was going to slow or stop to allow us to merge into the traffic, and that there would be sufficient space to do that.” Apparently, so did the car, right before it drove out in front of an oncoming bus. The bus driver didn’t expect the car to pull (slowly) into its lane, either.

All of the other self-driving car accidents to date have been the fault of other drivers, and the authors think this is telling. If you unexpectedly brake all the time, you can probably expect to eventually get hit from behind. If people can’t read your car’s AI’s mind, you’re gonna get your fender bent.

The paper’s solution is to make autonomous vehicles more predictable, and they mention a number of obvious solutions, from “I-sense-you” lights to inter-car communication. But then there are aspects we hadn’t thought about: specific markings that indicate the AIs capabilities, for instance. A cyclist signalling a left turn would really like to know if the car behind has the new bicyclist-handsignal-recognition upgrade before entering the lane. The ability to put your mind into the mind of the other car is crucial, and requires tons of information about the driver.

All of this may require and involve legislation. Intent and what all parties to an accident “should have known” are used in court to apportion blame in addition to the black-and-white of the law. When one of the parties is an AI, this gets murkier. How should you know what the algorithm should have been thinking? This is far from a solved problem, and it’s becoming more relevant.

We’ve written on the ethics of self-driving cars before, but simply in terms of their decision-making ability. This paper brings home the idea that we also need to be able to understand what they’re thinking, which is as much a human-interaction and legal problem as it is technological.

[Headline image: Google Self-Driving Car Project]