Off-road machine-learned obstacle navigation in an autonomous vehicle environment
An autonomous off-road vehicle, upon encountering an obstruction while navigating a route, can apply a first machine-learned model to identify the obstruction. In the event that the first machine-learned model cannot identify the obstruction, the autonomous off-road vehicle can apply a second machine-learned model configured to determine whether or not the obstruction can be ignored, for instance based on dimensions of the obstruction. If the obstruction can be ignored, the autonomous off-road vehicle can continue navigating the route. If the obstruction cannot be ignored, the autonomous off-road vehicle can modify the route, can stop, can flag the obstruction to a remote human operator, can modify an interface of a human operator to display a notification or a video feed from the vehicle, and the like.
1 . A method for autonomously navigating an autonomous off-road vehicle, comprising:
while autonomously navigating along a route, identifying, by the autonomous off-road vehicle, an obstruction obstructing the route;
determining, by the autonomous off-road vehicle, a confidence score representative of whether the obstruction can be ignored as the autonomous off-road vehicle navigates along the route such that if the confidence score is greater than a first threshold, the obstruction cannot be ignored and if the confidence score is less than a second threshold, the obstruction can be ignored; and
in response to the confidence score being less than the first threshold and greater than the second threshold, applying, by the autonomous off-road vehicle, a machine-learned model to sensor data representative of the obstruction, the machine-learned model configured to indicate whether the obstruction can be ignored or cannot be ignored.
2 . The method of claim 1 , wherein the confidence score is determined based on one or more of a type of the obstruction, a size of the obstruction, and dimensions of the obstruction.
3 . The method of claim 1 , wherein the machine-learned model comprises a neural network trained on previous interactions between a similar autonomous off-road vehicle and obstructions.
4 . The method of claim 1 , wherein the autonomous off-road vehicle is configured to perform an action based on the indication from the machine-learned model.
5 . The method of claim 4 , wherein the action performed by the autonomous off-road vehicle comprises one or more of: reducing a speed of the autonomous off-road vehicle, stopping the autonomous off-road vehicle, modifying the route the autonomous off-road vehicle is navigating, modifying an interface of a remote operator to indicate the obstruction, sending a notification indicating the obstruction to a device of the remote operator, and emitting a warning noise to the remote operator indicative of the obstruction.
6 . The method of claim 1 , wherein the machine-learned model is trained based on a set of training data representative of interactions between vehicles and obstructions, wherein the set of training data includes information describing a change in a condition of one or both of a vehicle and an obstruction with which the vehicle interacted.
7 . The method of claim 1 , wherein the sensor data is depth data captured by a depth-sensing system of the autonomous off-road vehicle.
8 . An autonomous off-road vehicle comprising a hardware processor and a non-transitory computer-readable storage medium storing executable instructions that, when executed by the hardware processor, cause the autonomous off-road vehicle to perform steps comprising:
while autonomously navigating along a route, identifying an obstruction obstructing the route;
determining a confidence score representative of whether the obstruction can be ignored as the autonomous off-road vehicle navigates along the route such that if the confidence score is greater than a first threshold, the obstruction cannot be ignored and if the confidence score is less than a second threshold, the obstruction can be ignored; and
in response to the confidence score being less than the first threshold and greater than the second threshold, applying a machine-learned model to sensor data representative of the obstruction, the machine-learned model configured to indicate whether the obstruction can be ignored or cannot be ignored.
9 . The autonomous off-road vehicle of claim 8 , wherein the confidence score is determined based on one or more of a type of the obstruction, a size of the obstruction, and dimensions of the obstruction.
10 . The autonomous off-road vehicle of claim 8 , wherein the machine-learned model comprises a neural network trained on previous interactions between a similar autonomous off-road vehicle and obstructions.
11 . The autonomous off-road vehicle of claim 8 , wherein the autonomous off-road vehicle is configured to perform an action based on the indication from the machine-learned model.
12 . The autonomous off-road vehicle of claim 11 , wherein the action performed by the autonomous off-road vehicle comprises one or more of: reducing a speed of the autonomous off-road vehicle, stopping the autonomous off-road vehicle, modifying the route the autonomous off-road vehicle is navigating, modifying an interface of a remote operator to indicate the obstruction, sending a notification indicating the obstruction to a device of the remote operator, and emitting a warning noise to the remote operator indicative of the obstruction.
13 . The autonomous off-road vehicle of claim 8 , wherein the machine-learned model is trained based on a set of training data representative of interactions between vehicles and obstructions, wherein the set of training data includes information describing a change in a condition of one or both of a vehicle and an obstruction with which the vehicle interacted.
14 . The autonomous off-road vehicle of claim 8 , wherein the sensor data is depth data captured by a depth-sensing system of the autonomous off-road vehicle.
15 . A non-transitory computer-readable storage medium storing executable instructions that, when executed by a hardware processor, cause an autonomous off-road vehicle to perform steps comprising:
while autonomously navigating along a route, identifying an obstruction obstructing the route;
determining a confidence score representative of whether the obstruction can be ignored as the autonomous off-road vehicle navigates along the route such that if the confidence score is greater than a first threshold, the obstruction cannot be ignored and if the confidence score is less than a second threshold, the obstruction can be ignored; and
in response to the confidence score being less than the first threshold and greater than the second threshold, applying a machine-learned model to sensor data representative of the obstruction, the machine-learned model configured to indicate whether the obstruction can be ignored or cannot be ignored.
16 . The non-transitory computer-readable storage medium of claim 15 , wherein the confidence score is determined based on one or more of a type of the obstruction, a size of the obstruction, and dimensions of the obstruction.
17 . The non-transitory computer-readable storage medium of claim 15 , wherein the machine-learned model comprises a neural network trained on previous interactions between a similar autonomous off-road vehicle and obstructions.
18 . The non-transitory computer-readable storage medium of claim 15 , wherein the instructions further cause the autonomous off-road vehicle to perform a step comprising:
performing an action based on the indication from the machine-learned model.
19 . The non-transitory computer-readable storage medium of claim 18 , wherein the action performed by the autonomous off-road vehicle comprises one or more of: reducing a speed of the autonomous off-road vehicle, stopping the autonomous off-road vehicle, modifying the route the autonomous off-road vehicle is navigating, modifying an interface of a remote operator to indicate the obstruction, sending a notification indicating the obstruction to a device of the remote operator, and emitting a warning noise to the remote operator indicative of the obstruction.
20 . The non-transitory computer-readable storage medium of claim 15 , wherein the machine-learned model is trained based on a set of training data representative of interactions between vehicles and obstructions, wherein the set of training data includes information describing a change in a condition of one or both of a vehicle and an obstruction with which the vehicle interacted.