Abstract: Systems and techniques are described for implementing autonomous control of powered earth-moving vehicles, including to automatically control movement of some or all of a powered earth-moving vehicle on a job site to conform with specified safety configuration data, such as to implement balancing of the vehicle(s) on non-level surfaces. For example, the safety configuration data may be used to move hydraulic arm(s) and/or attachment(s) of a vehicle while it is on a slope to prevent tipping or sliding, or to otherwise prevent moveable parts of a powered earth-moving vehicle (e.g., a rotatable chassis with a cabin; a tool attachment, such as a digging bucket, claw, hammer, blade, etc.; one or more hydraulic arms; etc.) from entering positions in three-dimensional (“3D”) space that are already occupied by other portions of the powered earth-moving vehicle (e.g., the chassis, tracks or wheels, etc.) and/or by other on-site obstacles.

Abstract: Systems and methods of bulldozer planning and control are disclosed.

Abstract: Systems and methods for using augmented learning models for autonomous earth-moving vehicles are disclosed. The method can comprise receiving a second set of sensor data; generating a first condensed vector from the second set of sensor data at least in part by processing the second set of sensor data with a first machine learning model; selecting an action to be performed by the vehicle at least in part by processing the first condensed vector with a second machine learning model. The method can further comprise retrieving one or more samples of sensor data from the first set of sensor data; fine-tuning the first machine learning model at least in part by processing the one or more samples of sensor data to produce a second condensed vector; and fine-tuning the second machine learning model at least in part by processing the second condensed vector.

Abstract: Systems and methods of bulldozer planning and control are disclosed.

Abstract: Systems and methods for using augmented learning models for autonomous earth-moving vehicles are disclosed. The method can comprise receiving a second set of sensor data; generating a first condensed vector from the second set of sensor data at least in part by processing the second set of sensor data with a first machine learning model; selecting an action to be performed by the vehicle at least in part by processing the first condensed vector with a second machine learning model. The method can further comprise retrieving one or more samples of sensor data from the first set of sensor data; fine-tuning the first machine learning model at least in part by processing the one or more samples of sensor data to produce a second condensed vector; and fine-tuning the second machine learning model at least in part by processing the second condensed vector.

Abstract: Systems and techniques are described for implementing autonomous control of powered earth-moving vehicles, including to automatically control movement of some or all of a powered earth-moving vehicle on a job site, such as to implement control of vehicle steering using a blade tool attachment in combination with vehicle tracks or wheels. For example, the autonomous operations may include monitoring an actual path or actual yaw direction that a powered earth-moving vehicle with a blade tool attachment (e.g., a bulldozer with a front blade tool) is following, and implement blade-based vehicle steering operations if the actual path or yaw direction differs from an intended target path (e.g., a straight path, a curved path, etc.) or target yaw direction, such as by lowering a side of the blade tool in a direction in which to cause an increase in vehicle turning.

Abstract: Provided herein are computer-implemented methods, systems, and media for operating a ripper attached to a machine based on an acceleration and one or more working parameters.

Abstract: Systems and methods of controlling an earth-moving vehicle (EMV) are disclosed. In one aspect, the system includes an EMV having a boom joint, a boom connected at the boom joint, a blade connected to an end of the boom, and a controller communicably coupled to the EMV, the boom and the blade. The controller is configured to move the EMV along a path, compute a target depth for the blade, position the blade to have the target depth, and dynamically adjust the blade to maintain the target depth as the EMV moves along the path.

Abstract: Systems and methods of bulldozer planning and control are disclosed.

Abstract: Systems and methods for using augmented learning models for autonomous earth-moving vehicles are disclosed. The method can comprise receiving a second set of sensor data; generating a first condensed vector from the second set of sensor data at least in part by processing the second set of sensor data with a first machine learning model; selecting an action to be performed by the vehicle at least in part by processing the first condensed vector with a second machine learning model. The method can further comprise retrieving one or more samples of sensor data from the first set of sensor data; fine-tuning the first machine learning model at least in part by processing the one or more samples of sensor data to produce a second condensed vector; and fine-tuning the second machine learning model at least in part by processing the second condensed vector.