Computer aboard the Mars rover may appear rudimentary, but it is truly special

The landing of Curiosity, National Aeronautics and Space Administration’s Mars rover, and the initial pictures from the surface of the fourth planet, have swelled the curiosity of people across the globe. The dangerous landing, dubbed the “seven minutes of terror”, is an unprecedented technological feat considering that the spacecraft carrying Curiosity had travelled more than 560 million km from Earth.

Curiosity has 17 cameras and a host of other equipment. It has chemical cameras, spectrometers, weather probes, radiation meters, rock drillers — literally a laboratory on aluminium wheels mounted on a titanium chassis. Remote-controlling the craft, which has been entrusted with the task of observing, analysing and reporting on the Martian surface and its ambience, is a major challenge, especially because of the large distance involved. Commanding the rover from here on a real-time basis is not feasible because the signal would take anywhere between 8 minutes and 28 minutes for a single round trip. Thus, Curiosity mainly relies on its own onboard intelligence, communicating back to the commanding station only once a day with the Jet Propulsion Laboratory of NASA.

Who does the thinking?

Interestingly, Curiosity’s computer is less powerful than the average Android-based smartphone available in market!

The motherboard (microcontroller, RAM, flash memory and other circuitry), RAD750, is the onboard computer for Curiosity and, on the lines of computation power, is humble when compared to the scale of terrestrial electronics available. It is clocked at 200 MHz (10 times faster than the previous Mars expedition, the Orbiter), has 256 MB of RAM and 2 GB flash to store the data it captures. By way of comparison, the average low-end Android smartphone available today is powered by nothing less than a processor clocked at 600 MHz, at least 256 MB RAM and supports for 32 GB of flash memory.

So, why does the $2.6 billion project rely on computational power generated by a device that is available in a sub-$100 smartphone?

Space-grade electronics is different from terrestrial electronics. Temperature and radiation effects can destroy electronics. For instance, the computer aboard Curiosity, the RAD750, can operate in temperatures ranging from –55 degrees Celsius to 125 degrees Celsius, certainly not the range in which a smartphone or a music player would be expected to operate.

While the atmosphere shields us from hazardous radiation emanating from the sun, high-energy radiation can ionise electronic components aboard Curiosity, which would lower performance. They are also vulnerable to ‘bit-flips’. Software programmes are stored on a computer as sequences of zeroes and ones (binary) in memory registers. Radiation can cause these bits to flip: when a zero becomes one, or a one becomes zero, the information contained in the programme is altered, rendering it useless.

The harsh conditions in space demand that space-grade electronics manufacturing processes enable them to be ruggedised, enabling them to cope with the ‘radiation hardening’, which impairs the computation power of the circuitry. The RAD750 is nothing but a radiation hardened version of IBM’s PowerPC750 and is the most powerful of processors available for space applications.

Real-time operation

However, these relatively low-end computers can handle highly computation-intensive tasks. It is just that the computer in space is not made to work many multiple tasks at a time. It toggles between mission critical tasks such as its own health monitoring and safe navigation and, maybe, switching on one or two payloads like the camera or the spectrometer. The coordination of tasks, and the real-time response to conditions the rover faces, is done by using a real-time operating system, commonly known as RTOS. Curiosity runs VxWorks by WindRiver, which is considered one of the most reliable RTOS available, and has flown aboard numerous space missions.

Constraints in space

“In a mission such as Curiosity, two of the major challenges are the limited bandwidth for communication and the time consumed by the cycle of ground analysis and subsequent commanding,” says Vigneswaran Karunanithi, a Master’s student at Technical University, Delft, who led the onboard computer team of India’s first pico satellite by students, STUDSAT, which was launched in 2010 by the Indian Space Research Organisation. “The rover cannot rely on the up-command from the earth to make all its decisions and has to make its own decisions when it comes to analysing data and prioritising onboard conditions,” Mr. Karunanithi says.

While operating through interplanetary distances, communication bandwidth and power pose major design constraints. To have ample bandwidth, signals must be transmitted at high frequencies, and in order to transmit signals at high frequency, more power is required.

On a 110-watt rover such as Curiosity, which is powered by the heat generated by the radioactive decay of plutonium dioxide, onboard power is still a precious commodity. This is one reason why the rover relies crucially on its own computer.

Moreover, with a flash memory of only 2 GB, and to perform numerous experiments and report back the data, the storage memory on Curiosity would exhaust within few tasks. This problem is solved by evacuating back to the commanding station on the Earth, either by transmitting high power signals using high gain antennas to the Deep Space Network (worldwide network of large antennas on the earth for interplanetary communication), or by handing over the data at low frequency and at lower power signals to the neighbouring Mars Reconnaissance Orbiter spacecraft that is orbiting Mars, which in turn transmits to the commanding station on the earth.

The case of Curiosity is a classic illustration of how raw computational power is not everything — even in the age of supercomputers.

(The author was one of the core team members of STUDSAT, and served on its onboard computer team for two years.)