Nilsson’s paper entitled “A Mobile Automaton: An Application Of Artificial Intelligence Techniques” discusses in detail the experiments conducted on a robot. The main goal of the project was the development of a system of programs for processing sensory data from the vehicle, for storing relevant information about the environment, and for planning the sequence of motor actions necessary to accomplish tasks in the environment. The paper discusses the research in the integration of the planning systems, models of the world and sensory processing systems into an “efficient whole capable of performing a wide range of tasks in a real environment”. In other words, study the processes for the real-time control of a robot system that interacts with a complex environment.
The paper itself was written in 1969, well before any of the latest architectures were conceived of. It is fascinating to actually read about how a mobile robot from 35+ years ago interacted with its environment, especially for problem solving. We can see from the paper that the world model was an important concept and that the robot needed to know everything about its environment in order to interact with it. This is very different from the more behavioral and response based mobile robots that we have today. The most fascinating section of the entire paper was the image processing of the environment. In first reading the paper, I was under the impression that the robot was mapping the environment. In a second pass, I realized that the robot was in fact given a planer map of the room which was called the “Grid Model”. It is interesting how the robot processed this map into a graph of nodes and utilized a shortest path algorithm to determine obstacles and how to travel to set goals.
The other section of this paper that was interesting was the “Visual Perception” section which spoke about how vision had the potential to be the most effective means for the robot system to obtain information about its world. I found this to be interesting because even today, vision presents numerous problems for mobile robots. Even though the resolution was limited in 1969, the algorithms for image processing seemed to be sophisticated for its time. By way of analyzing the different intensities at each point, the algorithm is able to decompose the image into line drawings for computation. Finally, one point that the article brings up is that in studying robot systems that interact with the real world, it is extremely important to build a program in a real system and to provide it with a real environment. This point continues today that a simulated environment does not present the variables and issues posed by a real environment.
Reference:
- Nilsson, Nils J. “A Mobile Automaton: An Application of Artificial Intelligence Techniques.” <http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.80.8608&rep=rep1&type=pdf>
