Method for setting parameters in an autonomous working device and an autonomous working device

1. A method for setting parameters in an autonomous working device, the autonomous working device being controlled based on a plurality of working parameters configured for the autonomous working device to perform a working task, the method comprising, in a process of generating working parameters:

generating for each of a plurality of different working environments a set of sensing values that sensors of the autonomous working device produce while being operated in the respective working environment;

partitioning the plurality of sets of sensing values into categories using a numerical clustering algorithm, determining for each cluster corresponding to one category, a representative set of sensing values, each category corresponding to one prototypical working environment;

optimizing the working parameters for each category to find an optimized working parameter set for each prototypical working environment;

storing the sets of optimized parameters in a memory of the autonomous working device, wherein each optimized working parameter set for each prototypical working environment is stored in association with a respective representative set of sensing values; wherein the method further comprises in the process of generating the working parameters:

generating for an individual working environment in which operation of the autonomous working device shall be performed an individual set of sensing values that the sensors of the autonomous working device produce while being operated in this individual working environment;

determining, based at least on the individual set of sensing values, among the prototypical working environments stored in the memory, the prototypical working environment showing highest similarity to the individual working environment;

setting the working parameters in the autonomous working device according to the optimized parameter set stored in the memory corresponding to the determined prototypical working environment; and

performing, by the autonomous working device, the working task based on the optimized parameter set.

2. The method according to claim 1 , wherein

generation of sensing values for the different working environments and the individual working environment is performed using a same predetermined set of working parameters.

3. The method according to claim 1 , wherein

a data compression is executed on the generated sensing values by transforming the generated sensing values into a low-dimensional representation for each working environment.

4. The method according to claim 1 , wherein

for determining the prototypical working environment with the highest similarity, a comparison of the individual set of sensing values and the representative set of sensing values is done.

5. The method according to claim 1 , wherein

each set of sensing values consists of a complete time series of sensing values or features derived therefrom during a fixed period of time.

6. The method according to claim 5 , wherein

a data compression is executed on the generated sensing values by transforming the generated sensing values into a low-dimensional representation for each working environment, and wherein

a histogram on events that are derived from the sensing values is generated.

7. The method according to claim 1 , wherein

the optimization is done with a nature inspired black-box optimization algorithm, in particular an evolutionary strategy algorithm.

8. The method according to claim 1 , wherein

the autonomous working device is an autonomous lawn mower.

9. The method according to claim 1 , wherein

the autonomous working device is an autonomous indoor cleaning robot.

10. The method according to claim 1 , wherein

the autonomous working device is an autonomous agricultural care robot or a harvesting robot or a seeding robot.