Back to results
US 11,238,594 B2Granted
Sidewalk edge finder system and method
AssigneeStarship Technologies Oü
InventorRasmus Raag
Life of this patent
Estimating the term…
Abstract
A method includes acquiring at least one image with at least one camera associated with at least one mobile robot; and extracting a plurality of straight lines from the at least one image; creating at least one dataset comprising data related to the plurality of straight lines extracted from the at least one image; forming a plurality of hypotheses for a walkway boundary based on the at least one dataset and determining at least one hypothesis with the highest likelihood of representing a walkway boundary; and using the at least one hypothesis to adjust a direction and/or speed of motion of the at least one mobile robot.
Claims26
A method comprising:
- (a) acquiring at least one image with at least one camera associated with at least one mobile robot;
- (b) extracting a plurality of lines from the at least one image acquired in (a);
- (c) creating at least one dataset comprising data related to the plurality of lines extracted from the at least one image in (b);
- (d) forming a plurality of hypotheses for a walkway boundary based on the at least one dataset and determining at least one hypothesis with a highest likelihood of representing a walkway boundary;
- and
- (e) using the at least one hypothesis determined in (d) to navigate the at least one mobile robot,
- wherein a line extracted in (b) is categorized as either a left line or a right line with respect to the at least one mobile robot;
- the method further comprising:
- (f) making a left partial hypothesis based on the lines to the left of the at least one mobile robot and a right partial hypothesis based on the lines to the right of the at least one mobile robot;
- (g) combining the partial hypotheses to make a full hypothesis, and
- (h) assigning a likelihood value to said full hypothesis,
- wherein the at least one dataset is formed by projecting the extracted lines on a certain projection plane, and wherein the certain projection plane comprises a ground plane defined with respect to the at least one mobile robot, and
- wherein, for each particular line projected onto the ground plane, two points are extracted, corresponding to endpoints of the particular line, and each of the partial hypothesis is based on a different collection of such points,
- wherein each of the partial hypothesis is formulated by applying a linear regression algorithm to a different collection of points;
- and
- the method further comprising:
- (i) assigning weights to the points for partial and full hypotheses likelihood calculation, the weights differing for at least two points,
- wherein the weights depend on a distance of the points from the at least one mobile robot.
15 more elements
A system comprising:
- (a) at least one camera associated with at least one mobile robot, the at least one camera adapted to acquire at least one image;
- and
- (b) at least one processing component constructed and adapted to at least extract a plurality of lines from the at least one image, create at least one dataset comprising data related to the plurality of lines extracted from the at least one image, and form a plurality of hypotheses for a walkway boundary based on the at least one dataset and determine at least one hypothesis with the highest likelihood of representing a walkway boundary;
- wherein each line is categorized as either a left line or a right line with respect to the at least one mobile robot,
- wherein the at least one processing component is further constructed and adapted to:
- make a left partial hypothesis based on the lines to the left of the at least one mobile robot and a right partial hypothesis based on the lines to the right of the at least one mobile robot;
- combine the partial hypotheses to make a full hypothesis;
- and
- assign a likelihood value to said full hypothesis,
- wherein the at least one dataset is formed by projecting the extracted lines on a certain projection plane, and wherein the certain projection plane comprises a ground plane defined with respect to the at least one mobile robot, and
- wherein for each particular line projected onto the ground plane, two points are extracted, preferably corresponding to endpoints of the particular line, and each of the partial hypothesis is based on a different collection of such points,
- wherein each of the partial hypothesis is formulated by applying a linear regression algorithm to a different collection of points,
- wherein the at least one processing component is further constructed and adapted to
- assign weights to the points for partial and full hypotheses likelihood calculation, the weights differing for at least two points,
- wherein the weights depend on a distance of the points from the at least one mobile robot.
12 more elements
A non-transitory computer-readable medium with one or more computer programs stored therein that, when executed by one or more processors, cause the one or more processors to perform at least the operations of:
- (a) acquiring at least one image with at least one camera associated with at least one mobile robot;
- and
- (b) extracting a plurality of lines from the at least one image acquired in (a);
- (c) creating at least one dataset comprising data related to the plurality of lines extracted from the at least one image in (b);
- (d) forming a plurality of hypotheses for a walkway boundary based on the at least one dataset and determining at least one hypothesis with the highest likelihood of representing a walkway boundary;
- and
- (e) using the at least one hypothesis determined in (d) to navigate the at least one mobile robot;
- wherein each line is categorized as either a left line or a right line with respect to the at least one mobile robot;
- (f) making a left partial hypothesis based on the lines to the left of the at least one mobile robot and a right partial hypothesis based on the lines to the right of the at least one mobile robot;
- (g) combining the partial hypotheses to make a full hypothesis;
- (h) assigning a likelihood value to said full hypothesis,
- wherein the at least one dataset is formed by projecting the extracted lines on a certain projection plane, and wherein the certain projection plane comprises a ground plane defined with respect to the at least one mobile robot, and
- wherein for each particular line projected onto the ground plane, two points are extracted, preferably corresponding to endpoints of the particular line, and each of the partial hypothesis is based on a different collection of such points,
- wherein each of the partial hypothesis is formulated by applying a linear regression algorithm to a different collection of points;
- and
- (i) assigning weights to the points for partial and full hypotheses likelihood calculation, the weights differing for at least two points,
- wherein the weights depend on a distance of the points from the at least one mobile robot.
14 more elements
Dependencies are inferred from the claim text.