Search Ames

Go

Feature

Text Size

Human and Robotic Exploration
02.27.07
 
robonaut QUESTION:
1) What do you see as the most significant challenges associated with the integration of human and robotic exploration in the future and what is the roadmap to achieve this vision? What role could NASA Ames Research Center play to realize this vision?

Left: Robonaut (foreground) performing a mock weld while Ames Research Center's K10 robot assists two EVA crew inspecting a previously welded seam.

ANSWER:
One central challenge in human and robotic exploration is the planning and scheduling of robotic activity. In many (and perhaps most) cases where humans and robots will work together to accomplish goals, the robots will not be entirely autonomous nor will they be joystickable. What will be required is for humans to very rapidly generate sophisticated activity plans for the robots they are working with. With its Scheduling and Planning Interface (SPIFe) and Europa intelligent scheduling system, Ames is now providing the robotic planning and scheduling systems for all current and upcoming Mars surface missions (MER '03, Phoenix '07, MSL '09). The work at Ames in this area is on the cutting edge of human-robotic interaction technology.

QUESTION:
2) From your vantage point what can human explorers do better than robots? What can robots do better than humans? What would, in your judgment, be an ideal division of responsibilities?

Right: Artist's concept of Orion.

ANSWER:
Orion Apollo had some automated systems, in particular for take-off and EDL (entry, descent and landing). It is clear that Orion (NASA's new spaceship capsule) will have more autonomy than Apollo. However, it may not be a lot more. Basically, in 2006 humans are still better at problem-solving and decision-making than machines, as was the case in 1966. As for autonomous robotic exploration of other planets, one of the primary requirements is sophisticated autonomous navigation. The current Mars Exploration Rover (MER) rovers, for example, do an impressive amount of autonomous navigation and hazard avoidance. From Earth, we basically send them a target and a rough path -- they solve the rest of the problems on their own, often for many meters. The challenge with planetary exploration is that we, the robot designers, don't have a very good understanding the environment in which the rover will have to perform (and hence why we are sending them in the first place). Mars, it turns out, is more forgiving in terms of dust and wind than had been anticipated, so the rovers have lasted almost four years now.

Artificial intelligence is often seen as key to the development of better robots. However, it is likely not just artificial intelligence, as traditionally defined, that will need to improve in order to allow autonomous exploration of other planets. It is also the robustness, adaptiveness and physical sophistication of our robotic capabilities. Much like the Segway (the two-wheeled, upright "human transporter" introduced by inventor Dean Kamen in 2001 ) provided a significant step forward in balancing capabilities, many such advances will need to be in place in order to support exploration of Europa (a moon of Jupiter), for example. The evolution of the opposable thumb in humans may have provided as much of an advantage to us as our ability to do math. Robotics, of the sort required to provide better vacuum cleaning of living rooms as well as better exploration of the solar system will evolve quickly over the next few decades. What we will have in 2030 will be more than adequate to support sophisticated, robust, autonomous exploration of the worlds around us. These systems may still not be able to independently come up with the theory of relativity, or even the Pythagorean Theorem though, as artificial intelligence (AI) researchers once hoped. Future missions will increasingly use robotic elements of all shapes and sizes. Just like artificial intelligence has evolved into efforts on narrowly focused intelligent systems (e.g., machine learning, reactive planners, machine vision, etc.) rather than broad intelligent agents, so too will robotics likely evolve into areas like robotic eyes, robotic drills, robotic vehicles, robotic and so on. Automated docking systems, self-deploying habitats, semi-autonomous exploration vehicles and other similar technologies that require some intelligent navigation, hazard-avoidance, targeting and even contingent decision-making will continue evolve as we work to return humans to moon.

Alonso Vera Alonso Vera, Ph.D.
Human-Systems Integration Division, NASA Ames

 
 
NASA Ames Research Center, Moffett Field, Calif.