Question: How long before driverless cars become commercially available?
The short answer is five years. The development of driverless cars is proceed incrementally, with cars already offering navigational aids and hands free parking. Carnegie Mellon is still developing driverless cars. This technology is also appearing in mining machines, construction machines, defense, agriculture, trucks, etc. These are all big money markets, and the R&D funding will only increase as a thriving industry develops.
Question: Roboticists are excited about the Kinect device. How will this change the field of robotics?
The Kinect device is an inexpensive device that can provide range data and sensing in light and dark environments. It is ideal for many different robotics applications. The sensing capabilities of the kinect have been around for a while, but they were prohibitively expensive for most robotic applications. But the kinect costs maybe $100 dollars, and Microsoft will soon introduce an improved version with greater sensitivity. So it has the potential to be a game-changer for robotics.
Question: How has funding for robotics research changed during the past decade?
Funding for robotics is now greater than it ever has been before. Twenty years ago, it would have been irresponsible for a major company to commit to robotics in mining, driving, defense, or virtually any field except manufacturing. The enabling technologies simply weren't ready. Now that the enabling technologies have been developed, companies are proliferating, new startups are frequently emerging, and big companies are adopting. So this is way beyond simply funding science projects.
Question: Carnegie Mellon is one of the contestants for the lunar x-prize. How confident are you that you will be able to land a rover on the moon by 2015?
We are very confident of landing a rover on the moon. There are currently 26 teams competing for the prize. The prize is $20 million for a non-federal program that lands a rover on the moon. Our ambition is to reach the pole to drill for lunar ice. We have an active program, and we are making steady progress.
Question: Are you concerned that the Chinese Change 3 lunar rover, set to land in 2013, might accomplish the task first?
The prize is for a non-federal moon landing. If a Government succeeds before a team, the prize is reduced to $15 million. I am amazed by the effectiveness of the x-prize to jump-start research. These prizes award one shot accomplishments, but the larger impact is to springboard further development.
Question: What role will the lunar X-prize have on robotics?
The cash prize is an incentive. A team will spend more than the prize to succeed, but the follow-on payoff is a multiple of the expenditure. This is why x-prizes are so effective, It provides an incentive to employ many different approaches. The robotic x-prizes are specifically designed to spur R&D on robotics applications, and they succeed at that.
Question: How important is Government funding to your project?
Government money for any of our projects is a component of a balanced portfolio. The Government didn't put any money into Lindbergh crossing the Atlantic, or to Deep Blue or Watson. So grants are good for incremental research, but I am more excited by transformational R&D. For these high-risk, high-reward tasks, corporations and x-prizes are often more effective than grants.
Question: How much AI research is Carnegie Mellon doing?
Carnegie Mellon essentially created the field of Artificial Intelligence. Carnegie Mellon currently has one of the best AI programs on the planet, and there is extensive collaboration between AI and robotics.
Question: How important are AI advances to robotic development?
Although AI advances do benefit the robotics industry, the robotics industry can grow exponentially even without embracing AI breakthroughs. For instance, our mining robot had to solve various issues involved with navigating in dark, uneven tunnels. Although this robot was limited in its capabilities, it was good enough to get the job done, and it succeeded in all of its objectives. That is the case with many robotic tasks - robots simply don't require the general, flexible, and subtle intelligence that humans have for most jobs they will be given.
Question: But what about robots that need to work outside of structured environments?
Unstructured environments do require more sophisticated AI, but even in those environments narrow AI programs generally get the job done. For instance, we developed a robot to drive around Antarctica, looking for meteorites. The robot was effectively able to navigate, and to locate meteorites. The same is true of the mars rovers - they can't play chess or Jeopardy, but they effectively perform the tasks that they are programmed to do. But for narrow AI tasks, robots often outperform humans.
Question: What robotic advances would you like to see by 2022?
By 2022, we should have made enough progress in areas such as sensing, planning, modeling, autonomy, and power, that the capabilities of robots will be approaching those of biological systems. We should see exponential growth both in the number of robots in operation and in the capabilities of robots. I am convinced that robotics will play as large a role in society during the next three decades as computers and the internet have played in the last three.
If you liked this article, please give it a quick review on ycombinator or StumbleUpon. Thanks