Aspects of the present disclosure involve systems, methods, and devices for mitigating Lidar cross-talk. Consistent with some embodiments, a Lidar system is configured to include one or more noise source detectors that detect noise signals that may produce noise in return signals received at the Lidar system. A noise source detector comprises a light sensor to receive a noise signal produced by a noise source and a timing circuit to provide a timing signal indicative of a direction of the noise source relative to an autonomous vehicle on which the Lidar system is mounted. A noise source may be an external Lidar system or a surface in the surrounding environment that is reflecting light signals such as those emitted by an external Lidar system.

Methods and systems for performing three-dimensional (3-D) LIDAR measurements with multiple illumination beams scanned over a 3-D environment are described herein. In one aspect, illumination light from each LIDAR measurement channel is emitted to the surrounding environment in a different direction by a beam scanning device. The beam scanning device also directs each amount of return measurement light onto a corresponding photodetector. In some embodiments, a beam scanning device includes a scanning mirror rotated in an oscillatory manner about an axis of rotation by an actuator in accordance with command signals generated by a master controller. In some embodiments, the light source and photodetector associated with each LIDAR measurement channel are moved in two dimensions relative to beam shaping optics employed to collimate light emitted from the light source. The relative motion causes the illumination beams to sweep over a range of the 3-D environment under measurement.

Methods and systems for performing three-dimensional (3-D) LIDAR measurements with multiple illumination beams scanned over a 3-D environment are described herein. In one aspect, illumination light from each LIDAR measurement channel is emitted to the surrounding environment in a different direction by a beam scanning device. The beam scanning device also directs each amount of return measurement light onto a corresponding photodetector. In some embodiments, a beam scanning device includes a scanning mirror rotated in an oscillatory manner about an axis of rotation by an actuator in accordance with command signals generated by a master controller. In some embodiments, the light source and photodetector associated with each LIDAR measurement channel are moved in two dimensions relative to beam shaping optics employed to collimate light emitted from the light source. The relative motion causes the illumination beams to sweep over a range of the 3-D environment under measurement.

A system configured to receive video and/or images from an image capture device over a livestock path, generate feature maps from an image of the video by applying at least a first convolutional neural network, slide a window across the feature maps to obtain a plurality of anchor shapes, determine if each anchor shape contains an object to generate a plurality of regions of interest, each of the plurality of regions of interest being a non-rectangular, polygonal shape, extract feature maps from each region of interest, classify objects in each region of interest, in parallel with classification, predict segmentation masks on at least a subset of the regions of interest in a pixel-to-pixel manner, identify individual animals within the objects based on classifications and the segmentation masks, and count individual animals based on identification, and provide the count to a digital device for display, processing, and/or reporting.

A system configured to receive video and/or images from an image capture device over a livestock path, generate feature maps from an image of the video by applying at least a first convolutional neural network, slide a window across the feature maps to obtain a plurality of anchor shapes, determine if each anchor shape contains an object to generate a plurality of regions of interest, each of the plurality of regions of interest being a non-rectangular, polygonal shape, extract feature maps from each region of interest, classify objects in each region of interest, in parallel with classification, predict segmentation masks on at least a subset of the regions of interest in a pixel-to-pixel manner, identify individual animals within the objects based on classifications and the segmentation masks, and count individual animals based on identification, and provide the count to a digital device for display, processing, and/or reporting.

A method includes: deriving a first absolute motion of the first optical sensor based on radial positions, azimuthal positions, radial distances, and radial velocities of points in a first cluster of points representing a first static reference surface in a first frame captured by the first optical sensor; deriving a second absolute motion of the second optical sensor based on radial positions, azimuthal positions, radial distances, and radial velocities of points in a first cluster of points representing a first static reference surface in a second frame captured by the second optical sensor; calculating a rotational offset between the first optical sensor and the second optical sensor based on the first absolute motion and the second absolute motion; and aligning a third frame captured by the first optical sensor with a fourth frame captured by the second optical sensor based on the rotational offset.

A scanning system includes a plurality of light detection and ranging (LiDAR) sensors, each LiDAR sensor of the plurality of LiDAR sensors being configured to generate a point cloud based on interactions of emitted light with a surrounding environment, and a processor configured to split a full 360° rotation of each LiDAR sensor into a plurality of slices including a first slice and a second slice, to generate a coherent fused point cloud by fusing together portions of point clouds of the plurality of LiDAR sensors corresponding to the first slice, and fusing together portions of point clouds of the plurality of LiDAR sensors corresponding to the second slice, to determine an action for the scanning system to take in response to the generation of the coherent fused point cloud, and to cause the scanning system to implement the action.

In one embodiment, a modular sensor assembly configured for mounting on a vehicle includes a first set of sensors and a second set of sensors. The modular sensor assembly includes a coordinate frame baseplate including a continuous surface, and sensor mounting elements coupled to the continuous surface for mounting the first set of sensors at a first height. The coordinate frame baseplate includes a sensor platform configured for mounting the second set of sensors at a second height. The first set of sensors and the second set of sensors are coupled to the coordinate frame baseplate so as to impart a common coordinate frame for the first set of sensors mounted at the first height and the second set of sensors mounted at the second height. The modular sensor assembly includes a bridging support structure coupled to the coordinate frame baseplate and capable of being mounted on a vehicle.

An autonomous articulated soil compactor machine can include: a machine frame; at least one cylindrical roller drum rotatably coupled to the machine frame and rotatable about a drum axis oriented generally transverse to a direction of travel of the compactor machine; a first lidar sensor on a front of the machine; a second lidar sensor on a first side of the machine; and a third lidar sensor on a second side of the machine; wherein the first, second and the third lidar sensors are positioned such that 360 degree lidar coverage is provided around the articulated compactor machine.

Systems, methods, and computer-readable media are provided for detecting whether an object is approaching an autonomous vehicle when a passenger is about to exit the autonomous vehicle, determining a speed and type of the object that is approaching the autonomous vehicle, determining whether the object will enter an exit zone when the passenger enters the exit zone based on the speed and the type of the object that is approaching the autonomous vehicle, and preventing the passenger from exiting the autonomous vehicle based on the determining of whether the object will enter the exit zone when the passenger enters the exit zone.