Planning system, planning method, and non-transitory computer readable storage medium

A planning method executed by a planning system including at least one processor, comprising:

• running a simulation to repeatedly generate candidate positions which are candidates for a target position with respect to a motion of a robot;
• selecting the target position among the candidate positions;
• and
• the running the simulation comprising:
• calculating an evaluation value corresponding to each of the candidate positions;
• generating a calculation model showing a relationship between each of the candidate positions and the evaluation value with a regression analysis;
• and
• setting a new candidate position based on the calculation model.

A non-transitory computer readable storage medium retrievably storing a computer-executable program therein, the computer-executable program causing a computer to perform a planning method, the planning method comprising:

• running a simulation to repeatedly generate candidate positions which are candidates for a target position with respect to a motion of a robot;
• selecting the target position among the candidate positions;
• and
• the running the simulation comprising:
• calculating an evaluation value corresponding to each of the candidate positions;
• generating a calculation model showing a relationship between each of the candidate positions and the evaluation value with a regression analysis;
• and
• setting a new candidate position based on the calculation model.

A planning system comprising:

• search circuitry configured to run a simulation to repeatedly set candidate values which are candidates for a target value;
• selection circuity configured to select the target value among the candidate values;
• and
• the search circuitry being configured to:
• calculate an evaluation value for an initial candidate value based on a first strategy;
• generate a first calculation model showing a relationship between the initial candidate value and the evaluation value with a regression analysis;
• set a plurality of initial candidate values based on the first calculation model;
• calculate a second evaluation value based on a second strategy having a higher accuracy than the first strategy for the candidate value set by using the plurality of initial candidate values;
• generate a second calculation model indicating a relationship between the set candidate value and the second evaluation value with the regression analysis;
• and
• set a new candidate value based on the second calculation model.