Space is hard. Anything from a stuck valve to a tiny mistake in a single line of code can turn a $100 million spacecraft into a pile of space dust. Not only does success or failure turn on a dime, but these factors are also affected by severe design constraints imposed by gravity, cosmic rays, and the merciless vacuum. In short, these conditions are perfect for testing designs that may one day end up on Earth.
A few months ago, we ran a blog on the four coolest NASA robots. We mentioned that some of the technologies seen in the Spirit, Opportunity, and Curiosity rovers looked like they could lend themselves to practical robotics applications such as self-driving cars. Now, we're going to shed a little more light on how that technology will make it from deep space to the interstate.
Design amidst Constraints
The biggest stumbling block on getting sophisticated robots to space is radiation. Outer space is flooded with extremely heavy atomic nuclei, moving almost as fast as the speed of light, which are collectively known as cosmic rays. There's also the occasional massive solar flare to consider. Only radiation-hardened electronic components can withstand the onslaught of solar and cosmic radiation. The problem with this is that radiation-hardened chips tend to be a few years behind the cutting edge of commercial microprocessors.
Secondly, there's the light speed delay. At closest approach, radio signals from Earth take about eight minutes to get to Mars. That delay precludes the idea of simply driving a Mars rover around like a remote-controlled car. Rather, the rover will need some autonomous capabilities just so it can perform basic tasks.
In other words, here are just a few paradoxical design constraints that you have to overcome if you want to get a robot into space:
- The robot needs to navigate the terrain of a hostile planet autonomously, due to radio delays.
- The practicalities of radiation shielding means that the hardware which processes autonomous movement will be several years out of date.
Why Are Limited Mars Robots Great for Cars on Earth?
Think about your car for a second. If it's a model that's been produced in the last couple of decades, it has a computer in it. Has that computer ever been upgraded or modified in the service lifetime of the car? If your car has a backup camera, have you ever swapped it out for one that produces higher-resolution images? The fact is, cars and NASA robots have one thing in common: they often feature low-powered, obsolete computing components.
Now, let's talk about autonomous cars. Cars can drive themselves using a combination of sensors and digital hardware. There's a LiDar package for detecting vehicles and pedestrians, and a software package that translates sensor data into driving instructions. While the cost of these separate components has been decreasing over the years, they still aren't what you'd call cheap—the autonomous upgrade for Tesla vehicles is projected to cost around $8,000.
Meanwhile, NASA has been funding research which would provide sophisticated image-recognition, navigation, and machine-learning capabilities, relying on technologies no more sophisticated that cheap digital cameras and low-powered cellphone processors. In other words, the same technology that's going to drive rovers on other planets may someday allow your cellphone to drive your car.
The best innovations in robotics are not always the ones that involve the most expensive, top-of-the-line computer hardware. The ones that catch on will probably use tried-and-tested technology, older than you would expect, but programmed for expanded capabilities. That philosophy serves NASA well—and it will definitely serve robots well here on Earth.