Abstract: In some examples, systems and methods are described for output biasing maneuvers recommendations provided by at least one machine learning maneuver-recommendation (MLM) engine executing on an aerial vehicle. In some examples, output biasing data can be received that includes at least one risk tuning parameter that can influence which of the maneuver recommendations are selected by a maneuver decision engine executing on the aerial vehicle based on a maneuver confidence threshold for implementation by the aerial vehicle. The maneuver confidence threshold can be updated based on the at least one risk tuning parameter to provide an updated maneuver confidence threshold for the output biasing of the maneuvers recommendation provided by the at least one MLM engine. Vehicle command data for implementing a given maneuver recommendation can be outputted based on an evaluation of the updated maneuver confidence threshold.

Abstract: Systems and methods are described herein for determining an optimized flight route for an aerial vehicle. In some examples, weather conditions for the aerial vehicle during a flight can be predicted based on weather data. At least two flight route segments based on the predicted weather data can be determined. The at least two flight route segments can include one of a solar flight route segment and a thermal flight route segment. A respective flight route segment of the at least two flight route segments can be discarded that can cause the aerial vehicle to violate a flight constraint. A replacement flight route segment for the respective discarded flight route segment can be determined. An optimized flight route for the aerial vehicle can be generated based on the replacement flight route segment and at least one remaining flight route segment of the at least two flight route segments.

Abstract: A machine learning maneuver model can be programmed to generate maneuver data identifying a plurality of flight paths for maneuvering an aerial vehicle through an adverse weather condition and a flight path confidence score for each flight path of the plurality of flight paths based on at least weather sensor data characterizing the adverse weather condition. The flight path confidence score can be indicative of a probability of successfully maneuvering the aerial vehicle through the adverse weather condition according to a respective flight path. A maneuver decision engine can be programmed to evaluate each flight path confidence score for each flight path relative to a flight path confidence threshold to identify a given flight path of the plurality of flight paths through the adverse weather condition that poses a least amount of structural risk to the aerial vehicle.

Abstract: Systems and methods are described herein for determining an optimized flight route for an aerial vehicle. In some examples, weather conditions for the aerial vehicle during a flight can be predicted based on weather data. At least two flight route segments based on the predicted weather data can be determined. The at least two flight route segments can include one of a solar flight route segment and a thermal flight route segment. A respective flight route segment of the at least two flight route segments can be discarded that can cause the aerial vehicle to violate a flight constraint. A replacement flight route segment for the respective discarded flight route segment can be determined. An optimized flight route for the aerial vehicle can be generated based on the replacement flight route segment and at least one remaining flight route segment of the at least two flight route segments.

Abstract: One example includes a system control architecture that includes local control systems to provide respective condition signals associated with situational awareness conditions of an associated system architecture. A central system controller receives the condition signals, generates a control scheme for control of operational aspects of the associated system architecture based on the condition signals, and generates control signals based on the control scheme. The control scheme defines contributions of each of the local control systems to a control authority of each of the operational aspects. Operational components provide mechanical control of each of the operational aspects of the associated system architecture in response to the respective control signals to implement the control scheme.

Abstract: A machine learning maneuver model can be programmed to generate maneuver data identifying a plurality of flight paths for maneuvering an aerial vehicle through an adverse weather condition and a flight path confidence score for each flight path of the plurality of flight paths based on at least weather sensor data characterizing the adverse weather condition. The flight path confidence score can be indicative of a probability of successfully maneuvering the aerial vehicle through the adverse weather condition according to a respective flight path. A maneuver decision engine can be programmed to evaluate each flight path confidence score for each flight path relative to a flight path confidence threshold to identify a given flight path of the plurality of flight paths through the adverse weather condition that poses a least amount of structural risk to the aerial vehicle.

Abstract: In some examples, systems and methods are described for output biasing maneuvers recommendations provided by at least one machine learning maneuver-recommendation (MLM) engine executing on an aerial vehicle. In some examples, output biasing data can be received that includes at least one risk tuning parameter that can influence which of the maneuver recommendations are selected by a maneuver decision engine executing on the aerial vehicle based on a maneuver confidence threshold for implementation by the aerial vehicle. The maneuver confidence threshold can be updated based on the at least one risk tuning parameter to provide an updated maneuver confidence threshold for the output biasing of the maneuvers recommendation provided by the at least one MLM engine. Vehicle command data for implementing a given maneuver recommendation can be outputted based on an evaluation of the updated maneuver confidence threshold.

Abstract: A computer implemented method is described herein for post-execution evaluation of a machine-learning (ML) algorithm. The method can include receiving a post-execution version of the ML algorithm having a plurality of behavioral states. The method can include generating behavior identification data identifying a given behavioral state from the plurality of behavioral states of the ML algorithm. The given behavioral state can correspond to a decision-making state of the ML algorithm that the ML algorithm learned during an execution of the ML algorithm. A graphical user interface (GUI) can be generated based on the behavior identification data that includes a behavior object characterizing the given behavioral state of the ML algorithm. Behavior evaluation data can be generated based on a user's interaction with the behavior object. A learning process of the ML algorithm can be altered for future execution of the ML algorithm based on the behavior evaluation data.

Abstract: One example includes a system control architecture that includes local control systems to provide respective condition signals associated with situational awareness conditions of an associated system architecture. A central system controller receives the condition signals, generates a control scheme for control of operational aspects of the associated system architecture based on the condition signals, and generates control signals based on the control scheme. The control scheme defines contributions of each of the local control systems to a control authority of each of the operational aspects. Operational components provide mechanical control of each of the operational aspects of the associated system architecture in response to the respective control signals to implement the control scheme.