Robot brains to design the future?
Intelligent robots must analyse their worlds at many levels of abstraction. Human beings create new ways of looking at problems, and often their solutions relate to different problems to those first formulated. Planning involves identifying future possible states, selecting one or more as goals, devising trajectories to reach the goals, and devising control strategies to keep the system on trajectory. The future can be considered to be a huge space of possible worlds, and planning can be considered to be a search problem with the objective of finding relatively good future states. This idea of finding the future is passive and contrasts radically with the reality that human beings actively construct the future – we build the future as it ought to be and according to how we can make it come into being. Human beings create artificial systems, and the process of doing this is design. Designers do not solve problems using algorithms. They engage in a computationally irreducible generate-test process that accumulates information and understanding as it progresses. It is widely accepted in the design research community that there is a designerly way of solving problems, and that the process of doing design results in new ways of looking at problems, reframing them and producing emergent solutions that satisfice the many competing constraints. In human systems the future is what we make it and a defining feature of human intelligence is that we design the future. This presentation is based on recent experimental work indicating that solving open-ended design problems involves different ways of using of the human brain compared to solving logical or mathematical problems. This suggests that the architecture of robot brains can be fundamentally rethought from the perspective of design with planning and execution as the process of designing the future.
Jeffrey Johnson
国际会议
2008 Sino-European Workshop on Intelligent Robots and Systems(SEIROS08)(第一届中欧智能系统及机器人国际学术研讨会)
重庆
英文
1-1
2008-12-11（万方平台首次上网日期，不代表论文的发表时间）