Sunday 17 August 2014

Vision for Mars Rovers


Seeing is believing. Our great visions of space exploration require also a trustable vision system. Curiosity rover snapping dozens of pictures every day is capable to see the Red Planet in different way than the human eye. Needless to say, human vision is highly adapted to the specific conditions here on Earth. “The exploration of Mars will require radically different qualities of the vision system because of both the Mars atmospheric properties and the range of things that this vision system will need to be able to see.” Yosef Akhtman of the Swiss Federal Institute of Technology in Lausanne (EPFL) told astrowatch.net. “In particular using an RGB camera, which was developed to mimic human vision on Earth is a bit naive and far from optimum.” Akhtman will give a speech on Sept. 7 regarding adaptive vision system for extraterrestrial exploration at the European Mars Conference (EMC) that will take place in Podzamcze, Poland.

Akhtman is the founder and the Chief Executive Officer of VISNX – an innovative Swiss start-up company in the field of optical sensing. He is the author of several inventions, one manuscript and more than forty scientific papers, spanning the subjects of information theory, mobile robotics and remote environmental sensing.

Akhtman points out that vision, and the sensation of color in particular, is a strictly subjective, contextual capability that has evolved over billions of year to serve the needs of the various species in a very effective, but highly specialized manner. Bees have their vision system optimized for navigation and orientation around flowers. The vision of many mammals is optimized for the sensing of motion at the expense of the less resolved color information. Birds see more distinctive colors than humans, which is probably optimized for their airborne lifestyle. Yet another species like the mantis shrimp perceive up to twelve distinctive spectral bands.

Many of today's imaging technologies are designed to replicate the capabilities of the human vision. The goal of these systems is to record color information in a way that would facilitate faithful reproduction ­ “real color” that would resemble the original object when observed by humans.

“We don't yet know what is it exactly that we will need to look for on Mars and therefore we cannot develop an effective static vision system in advance. That is the reason that the adaptive approach that we are developing at VISNX has so much potential in my view.” Akhtman said. “We have developed a fundamentally different vision system that can be dynamically reconfigured to collect imaging data that is optimized for the specific environment. In particular, the spectral characteristics of the proposed imaging system may be adjusted in post processing to maximize the informational content of data collected.”

He added that the proposed vision system opens new and compelling possibilities for optical sensing both on Earth and other planets with the exploration of Mars being the clear and straightforward application.

Along with the EMC, an international Mars rover contest will be held - the European Rover Challenge. The challenge involving analogues of Mars rovers is a competition for teams of students and recent graduates of higher education institutions, who, with the help of their faculty, try to first design and build and then field the best rover. The core of the challenge are four practical tasks: a science task involving obtaining and analyzing samples, a “blind” navigation task, in which the team will have to guide the rover to a certain destination using just GPS coordinates and no camera input, and two engineering tasks that will require using and repairing equipment. During all of the tasks, the teams will have to control their rovers without seeing them directly.

Asked about what will be the most important to win the competition, Akhtman, who was the main developer of the control software for the Southampton Autonomous Underwater Vehicle, which took the 1st place at the SAUC-E competition in 2007 (sauc-europe.org), said: “I think flexibility and robustness of both hardware and software is very important. I am a big believer of redundant, asynchronous control architecture. These days you have wonderful platforms, such as ROS (ros.org) that are almost plug-n-play, so you don't need to develop such software from scratch the way we did in 2007. From my experience in these types of challenges, it is very common for the entire system to fail because of an expected failure of a small, but critical component. The challenge for the developers is to think through and eliminate any potential weak links and failure points.”

Astrowatch.net is the official media patron of the European Rover Challenge competition and the European Mars Conference.

No comments:

Post a Comment