3
$\begingroup$

According to the research paper Parallel process decomposition of a dynamic manipulation task: robotic sewing (D. Gershon, DOI: 10.1109/70.56654):

Abstract - ... The task planner approach, as promoted by the Al community, is unsuited to tasks involving interaction with a dynamic environment. ....
... ... ...
Task planning is essentially an off-line activity, based on a "snap-shot" of the world, and is therefore incompati- ble with dynamic tasks and environments. Task planners re- quire a single model of the world, whereas different models may be appropriate for different objectives, e.g., an octree representation is efficient for obstacle avoidance, whereas a RAPT-style model [36] is better suited to planning compliant motion tasks. Attempts to develop experimental task planner systems revealed additional difficulties, such as image under- standing, sensor fusion, error recovery, and the potential for catastrophic failures [29].... ..... ..... ....

Many control schemes, such as adaptive control [15], [27] and sliding-mode control [41], can accommodate bounded uncertainties in the model of the controlled system. However, these controllers are not generally robust to disturbances generated by a dynamic interacting environment.

Now, Multi-arm robot control system for manipulation of flexible materials in sewing operation seems to be implementing a multi-arm robot system on the basis of "Task Planning" approach. And, this paper implemented a robot system using adaptive control.

So, what is the catch here? Why does the 1st article say that use of task planning and adaptive control aren't possible in case of dynamic tasks?

$\endgroup$
1

1 Answer 1

3
$\begingroup$

I think what you are seeing here is a decade advancement in microprocessor and robotic control technologies.

By the time the second and third papers were written, in 2000 & 1998, the definition of 'task planning' had switched from static pre-planning to dynamic reactive planning.

The difference in microcomputer speeds between 1990 and 1998 is enormous. In 1990 typical CPUs included the 80486 & i860. They were clocked at double digit MHz, and didn't have the power to do the complex floating point operations required for doing task planning in real time.

By the late 90's microprocessors were significantly faster, GHz chips were going into production and CPUs were doing more per cycle. Significantly, SIMD (Single Instruction Multiple Data) pipelines like SSE were being implemented so the big matrix manipulations needed to do dynamic task planning were starting to become efficient enough to use in near real time.

This all made enough difference that suddenly things which were inconceivable at the start of the decade were now viable.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.