Artboard 3 Created with Sketch.
Snapshot

We need self-learning robots to take human civilisation to its next — and inevitable — stage.

When I was a child, we used to spend Durga Puja at the foot of the Ayodhya Hills, Purulia, where one of the fascinations was to spot artificial satellites, with our naked eyes, as they darted through the stars across the night sky. This Deepawali, parked in the pristine gardens of the Tumsong Tea Retreat near the dreamy town of Ghoom, Darjeeling, I was looking forward to a similar sky show when a thought struck me. Chinese fireworks may be out of favour this year, but right above me in the sky was Tiangong 2, the Chinese permanent space station where two Chinese taikonauts had just arrived for their month-long tour of duty.

Living in space is tough and the technology necessary to make space habitable and safe for humans is very difficult and expensive to develop. Hence China’s Taikonauts@Tiangong is an impressive feat, but is it really necessary? Should India try to catch up? Or is there an alternate, inexpensive approach? The sci-fi stories of Isaac Asimov’s famous “positronic” robots, that assist man in exploring space, suggest an elegant solution.

Given the degradation and deterioration of the terrestrial environment, it is but inevitable that humanity will eventually escape into the vastness of space. But with our current knowledge of physics and state of rocket technology, even nearby stars in the galaxy are simply too far to reach in a single human lifespan. So our search for habitable places is restricted to four principal candidates in our solar system. Mars is a close cousin of Earth in terms of terrain but has a very thin atmosphere. Titan has a dense atmosphere of nitrogen and methane and since it rains methane, it has huge hydrocarbon lakes. Both Enceladus and Europa have huge oceans of good, life-friendly liquid water and the latter has a thin oxygen atmosphere as well. All these are potential places where humans can plan on setting up colonies but unfortunately none of them will natively support life as we know it on Earth. The limitations of our terrestrial biology makes us unfit to survive in almost every space location. Our fragile human bodies have to be covered in either space suits or we have to create closed habitats that mimic our Earth in terms of temperature and atmosphere. How can we create such habitats that are big enough for a sustainable human population?

Given the astronomical amounts, pun intended, of energy required to lift material against our gravity and propelling them across space, it is clearly infeasible to have these habitats built on Earth and transported to their destinations. We have to look for ways to build them at the destination using materials that are sourced locally. Moreover, using human labour for construction leads us to a chicken-and-egg situation, because for humans to live and work safely in biologically hostile locations, the location has to be first made habitable. This is where robots can step in. But these need not be anthropomorphic or humanoid robots that we come across in popular imagination—the ones with blinking eyes that can walk or dance on two legs. Instead, they would be the kind of industrial robots that are already in widespread use in many automated factories today.

The first kind of robot that we would need would be like Google’s autonomous vehicles that can navigate across the terrain either on wheels or on tracks. Alternatively they could be drones that fly through whatever atmosphere that is available or autonomous submersibles that could explore the methane lakes on Titan or the ocean depths that lie under the ice sheets of Europa and Enceladus. These robots would be the pioneers, the pathfinders that will explore the terrain and search for locations where we may find the right kind of materials necessary for building the habitats.

The second kind of robots are the ones that can drill into, excavate, or otherwise collect material and transport them to a central fabrication location. Basically these would be advanced versions of current mining and material handling technology that is already in use or is being tried in autonomous mining operations in many countries today.

The third category of robots would be the kind that are used in automated factories to build or assemble components—components of living quarters, of energy sources, of chemical plants, of material transportation systems and of the factories themselves and its robots. In fact, robots must build other, simpler robots. All this manufacturing would in turn have to be done using either traditional material forming processes to create desired shapes or with 3D printers that are modified to work with the kind of materials that are excavated at the site. The 3D-printed concrete castle built by Andrey Rudenko is a step in this direction.

In fact, the technology for building these robots is well understood. But in space, they need modifications. First they need to operate under different environmental conditions—low gravity, bitter sub-zero temperatures and a very different chemical environment. Secondly they need to ensure compatibility and interoperability between available material, the process of its extraction, transportation, processing and subsequent assembly into products. But what is most important is reliability. On Earth, equipment can be repaired or replaced but in space that is not possible because that will take years. Since faults are inevitable, these robots must be built with redundancies and self-healing capabilities without increasing the weight. That is the real challenge.

These robots would also require an astonishing level of artificial intelligence to be able to operate almost autonomously. Asimov had playfully referred to the robots having “positronic” brains to distinguish them from simple electronic circuits that were then popular. Today, when hyper-exponential growth of deep learning and cognitive computing technologies is about to make digital intelligence more powerful than biological intelligence, it is important that these technologies be integrated into the next generation of extraterrestrial robots.

While the difficulty, and expense, of building, transporting and deploying these robots is high, they are several orders of magnitude lesser than what would be encountered if we send human beings. This is because a human being would need, in addition to all this capability, its own complex life support system along with redundant backups to address the inevitable accidents and failures. We can write off the expenses of a robot— a piece of machinery — if things go wrong, but the moral hazards of allowing a crew member to die in space because of a technical fault are far higher. So eliminating a human crew and replacing them with a crew of autonomous robots will lead to a dramatic reduction in costs and hence an equally dramatic increase in the economic feasibility of any such project.

One of the lingering tragedies of the Apollo programme that successfully landed men on the moon six times between 1969 and 1972, was that while it demonstrated man’s ability to arrive, it failed to initiate the process of building habitats. In the 45 years since man last walked and worked on the moon, even if a single habitable lunar colony was built, we would have had acquired by now the confidence for building habitats on Mars or Titan. Unfortunately, the focus then, and perhaps even today, is to be able to simply go there and claim bragging rights.

What we also need is to focus on what to do once we are there. Unlike tourists ticking off important destinations, we should plan to be immigrants who create homesteads and economic assets that will not only ease pressure on Earth but also serve as stepping stones to other destinations.

Today, NASA, ESA, ISRO and private entrepreneurs like Elon Musk, Jeff Bezos and Team Indus are racing each other to build the technology to travel into deep space. This technology is based on the efficient conversion of energy and smart control of very complex systems. There is no doubt that this is absolutely necessary if mankind has to liberate itself from an Earthbound existence and become a space faring species. But we also need another set of entrepreneurs who will focus on the complementary technology necessary to create a non-human construction crew that builds habitats. This technology will be based on robotics hardware that is controlled by artificial intelligence (AI) software — the technology of autonomous robots.

Team Indus is perhaps the only company in India that plans to go to the Moon by 2018 as part of the Google X-Prize competition and it would be nice if many more Indian companies were to create robots that ride on their craft and carry out mining and fabrication experiments on the Moon.

Fortunately, terrestrial business opportunities have created a big market for robotics and AI and today, instead of teaching machines how to perform tasks, we teach them to learn on their own how to perform these tasks. We need to adapt these technologies for use in space. That will be the grand realisation of Asimov’s fictional robots and a gigantic step for man’s journey towards the stars.