Navigation system based on slow feature gradients

Patent metadata

Field	Value
Publication number	US-10955852-B2
Application number	US-201815905962-A
Country	US
Kind code	B2
Filing date	Feb 27, 2018
Priority date	Feb 28, 2017
Publication date	Mar 23, 2021
Grant date	Mar 23, 2021

Abstract

A method and system for navigating a mobile system and corresponding mobile system is disclosed, in particular for autonomous mobile systems such as robots, for example lawn mowers or even smartphones. The mobile device includes at least one sensor, an electronic control unit, and an output unit. The method includes acquiring sensor data on an environment of the mobile device, calculating a gradient of a difference of a target environmental representation and a current environmental representation, and determining a movement direction to reach a target position corresponding to the target environmental representation based on the estimated gradient. The determined movement direction for navigating the mobile device is output, for example to a steering system of the mobile device or to a display. The method can include generating an environmental representation by performing unsupervised learning from the acquired sensor data.

First claim

The invention claimed is:

1. A method for navigating a mobile device, the mobile device comprising at least one sensor, an electronic control unit and an output unit, the method comprising:
acquiring sensor data from the at least one sensor on an environment of the mobile device;
generating environmental representations, in a training phase, from the acquired sensor data;
calculating, in a navigation phase, a gradient of a difference of a target environmental representation and a current environmental representation,
wherein the target environmental representation corresponds to a target location, and the current environmental representation corresponds to a current location of the mobile device, and are selected from the environmental representations generated in the training phase;
determining a movement direction for the mobile device to reach the target location corresponding to the target environmental representation based on the gradient; and
outputting the movement direction for navigating the mobile device.

2. The method for navigating a mobile device according to claim 1 , further comprising:
generating the environmental representations by performing unsupervised learning from the acquired sensor data.

3. The method for navigating a mobile device according to claim 2 , wherein
in the generating the environmental representations, a slowness learning rule is applied as the unsupervised learning.

4. The method for navigating a mobile device according to claim 2 , wherein
in the generating the environmental representations, a slow feature analysis is applied as the unsupervised learning.

5. The method for navigating a mobile device according to claim 1 , wherein
in the calculating the gradient, the gradient is calculated by fitting a plane to the difference.

6. The method for navigating a mobile device according to claim 5 , wherein
in the calculating the gradient, the gradient is calculated by fitting a plane through more than three points of the difference.

7. The method for navigating a mobile device according to claim 1 , wherein
in the calculating the gradient, the gradient is calculated for multiple dimensions of the difference, in particular by using a weighted sum of the multiple dimensions of the difference.

8. The method for navigating a mobile device according to claim 1 , wherein
in the calculating the gradient, the gradient is calculated for multiple dimensions of the difference by using a weighted sum of the multiple dimensions of the difference, wherein a measure of slowness is used for weighting.

9. The method for navigating a mobile device according to claim 1 , further comprising:
computing a plausibility of the environmental representations or of an estimated path of the mobile device by determining a rate of change of the environmental representations or the estimated path, and
initiating an action if the rate of change exceeds a predetermined threshold.

10. A computer program embodied on a non-transitory computer-readable medium, said computer-readable medium containing program-code which, when executed on a computer or digital signal processor, control the computer or digital signal processor to perform the method of claim 1 .

11. The method for navigating a mobile device according to claim 1 , wherein
in the outputting of the movement direction, an action is initiated if a rate of change exceeds a predetermined threshold, the action comprising at least one of issuing a warning to a user, repeating the steps of acquiring sensor data on an environment of the mobile device and generating an environmental representation from the acquired sensor data.

12. The method for navigating a mobile device according to claim 1 , further comprising:
configuring a movement trajectory of the mobile device for optimizing the calculating the gradient.

13. The method for navigating a mobile device according to claim 1 , further comprising:
repeating the calculating the gradient, determining the movement direction to reach the target position corresponding to the target environmental representation based on the gradient, and outputting the movement direction, until the difference is below a predetermined threshold, and
changing to a further navigation method when the difference is below the predetermined threshold.

14. A system for navigating a mobile device, the system comprising:
an electronic control unit connected with a memory and an output unit, and
an acquisition unit for acquiring sensor data on an environment of the mobile device from at least one sensor,
wherein the electronic control unit is configured to
generate, in a training phase, environmental representations from the acquired sensor data,
calculate, in a navigation phase, a gradient of a difference of a target environmental representation and a current environmental representation,
wherein the target environmental representation corresponds to a target location, and the current environmental representation corresponds to a current location of the mobile device, and are selected from the environmental representations generated in the training phase,
determine a movement direction for the mobile device to reach the target location corresponding to the target environmental representation based on the gradient, and
the output unit is configured to output the determined movement direction for navigating the mobile device.

15. The system for navigating a mobile device according to claim 14 , wherein
the mobile device comprises a smartphone or a vehicle, in particular an autonomously operating vehicle, or a robot, in particular a cleaning robot or an autonomous lawn mower.

16. The system for navigating a mobile device according to claim 14 , wherein
the system comprises plural sensors of a same sensor type arranged at spatially different sensor positions,
wherein the sensor positions of the plural sensors are known relative to each other, and
wherein the electronic control unit is configured to calculate the gradient by fitting a plane to the difference.

17. The system for navigating a mobile device according to claim 14 , wherein
the electronic control unit is configured to calculate the gradient further taking sensor data from past sensor readings into account.

18. A method for navigating a mobile device, the mobile device comprising at least one sensor, an electronic control unit and an output unit, the method comprising:
acquiring sensor data from the at least one sensor on an environment of the mobile device,
calculating a gradient of a difference of a target environmental representation and a current environmental representation,
determining a movement direction to reach a target position corresponding to the target environmental representation based on the gradient,
wherein the movement direction is modified to deviate from a direction of a fastest gradient descent, in particular to avoid collinear steps in a next method cycle, and
outputting the determined movement direction for navigating the mobile device.

19. The method for navigating a mobile device according to claim 18 , wherein
in the determining a movement direction, a momentum term is added or a simultaneous perturbation stochastic approximation approach is used.

20. A system for navigating a mobile device, the system comprising:
an electronic control unit connected with a memory and an output unit, and
an acquisition unit for acquiring sensor data on an environment of the mobile device from at least one sensor,
whe

Assignees

• Honda Res Inst Europe Gmbh

Inventors

• Franzius Mathias
• Metka Benjamin
• Bauer-Wersing Ute

Classifications

• G01C21/20 Primary

  Instruments for performing navigational calculations (G01C21/24, G01C21/26 take precedence) · CPC title

• G05D1/0221 Primary

  involving a learning process · CPC title

• G05D1/243 Primary

  Means capturing signals occurring naturally from the environment, e.g. ambient optical, acoustic, gravitational or magnetic signals (using passive navigation aids external to the vehicle G05D1/244; using signals from positioning sensors located off-board the vehicle G05D1/249) · CPC title