Abstract: A computing system establishes a geofence associated with a particular service area. The system monitors a location of a computing device of a driver and detects when the driver enters the geofence. The system places the driver into a queue for the particular service area. The system receives a transport request from a computing device of a requesting user, where the transport request is associated with the particular service area. The system arranges the transport service for the requesting user in accordance with the set of rules by (i) selecting a driver from the queue to service the transport request for the requesting user, and (ii) communicating at least one of: driver information of the selected driver, an estimated time of arrival to a pickup location, or a location of the selected driver to the computing device of the requesting user.

Abstract: Systems and methods for detecting and orchestrating the deployment of software changes. The system can access schema data indicative of a change associated with a first system, wherein executable code associated with the change is executable within a first computing environment. The system can determine a potential impact to a second system. The system can generate one or more computing tasks to notify the second system of the change and an acknowledgment placeholder with a unique identifier associated with the second system. The system can access data indicative of an execution status of the one or more computing tasks. The system can generate an update indicating that the one or more computing tasks have been executed by the second system. The system can transmit to the first system, data acknowledging the potential impact to the second system, and command instructions to deploy the executable code to a second computing environment.

Abstract: A transportation management system generates routing guidance from an origin location to a destination location by modifying edge weights in a graph of a geographic location to penalize difficult immediate maneuvers. Responsive to receiving a routing request, the system identifies a position of a provider device in a base map having edges representing road segments and nodes representing intersections between road segments. A sub-graph is generated for the edges in the base graph located up to a threshold distance from the origin location, and the system modifies the weight of one or more edges in the sub-graph corresponding to a difficult immediate maneuver. When applying a routing algorithm to generate the routing guidance, the system uses the edge weights of the generated sub-graph for a first portion of the routing guidance and the original edge weights of the base graph for a second portion of the routing guidance.

Abstract: Systems and methods for reconciling location based on multiple computing device signals. For example, the computing system can obtain location datasets associated with freight carrier services from computing sources. The computing system can determine an expected signal pattern for a location associated with a freight transportation service. The computing system can determine, for each computing source, a confidence score. The confidence score can represent the probability that the respective location dataset is associated with a load being transported for a freight transportation service. The computing system can determine a primary location dataset based on the confidence scores. The computing system can perform actions based on the primary location dataset.

Abstract: A method for user-initiated end of delivery confirmation. A method and technique comprises causing display of a first interactive element on a user device in response to a vehicle carrying a payload approaching a destination; receiving a first user input comprising a first user interaction with the first interactive element; in response to receiving the first user input, causing a vehicle management component communicatively coupled to the vehicle to enable access to the payload in the vehicle and causing display of a second interactive element on the user device; receiving a second user input from the user device, the second user input comprising a second user interaction with the second interactive element; determining a retrieval status associated with the payload based on the second user interaction; and causing display of a confirmation on the user device based on retrieval status, the confirmation indicating that the payload has been retrieved.

Abstract: A machine learning model for reinforcement learning uses parameterized families of Markov decision processes (MDP) with latent variables. The system uses latent variables to improve ability of models to transfer knowledge and generalize to new tasks. Accordingly, trained machine learning based models are able to work in unseen environments or combinations of conditions/factors that the machine learning model was never trained on. For example, robots or self-driving vehicles based on the machine learning based models are robust to changing goals and are able to adapt to novel reward functions or tasks flexibly while being able to transfer knowledge about environments and agents to new tasks.

Abstract: A network system determines dissimilarities between a digital map and trace data of a road network in an area as service providers and service requestors coordinate service using the road network in the area. To determine dissimilarities the network system can determine a suggested route, determine a predicted route, receive executed trace data, and compare the predicted route data to the executed trace data for the suggested route. The network system may aggregate trace data when determining a dissimilarity. The network system can quantify the differences between traces to determine dissimilarities. Quantifications can include, ratios, bounds, and scores. The network system can determine and alternate route if a dissimilarity indicates that the state of a road segment has changed (e.g., from “open” to “closed”). The network system can modify guidance instructions if a dissimilarity indicates that a guidance instruction is misleading.

Abstract: A network system obtains position information from a plurality of computing devices that are located within a geographic region where a mass-user facility is provided. A congestion level is determined for an area of the geographic region, based at least in part on the position information obtained from the plurality of computing devices. In response to determining that the congestion level exceeds a threshold level, the network system determines a target distribution profile for managing the use of multiple egress points by users within the geographic region to reduce the congestion level. Application content data can be transmitted to individual computing devices to cause corresponding users to use a corresponding of the multiple egress points, in accordance with the target distribution profile.

Abstract: Systems and methods for displaying corresponding content for vehicle services using a distributed set of electronic devices are provided. For example, a computer-implemented method includes obtaining data associated with a vehicle service instance. The vehicle service instance is associated with a request for a vehicle service for a user. The method includes determining, based on the data associated with the vehicle service instance, a first advertisement content item for a display device positioned on an exterior of a vehicle assigned to the vehicle service instance and a second advertisement content item for a user device associated with the vehicle service instance. The method includes communicating data that initiates the display of the first advertisement content item for the display device positioned on the exterior of the vehicle and data that initiates the display of the second advertisement content item for the user device.

Abstract: Computationally implemented methods and systems that are designed for receiving a request for transporting one or more end users towards a destination location; providing a travel plan for facilitating the one or more end users to travel to the destination location from a starting location, the travel plan identifying at least two route legs including at least one transport route leg that calls for at least one transportation vehicle unit to transport the one or more end users over the transport route leg; and directing the at least one transportation vehicle unit to rendezvous with the one or more end users at a rendezvous location in order to transport the one or more end users over the transport route leg. In addition to the foregoing, other aspects are described in the claims, drawings, and text.

Abstract: Systems and methods are disclosed herein for training neural networks that can be adapted to new inputs, new tasks, new environment, etc. by re-training them efficiently. A parameter vector is initialized for a neural network. Perturbed parameter vectors are determined using the parameter vector. Behavior characteristics are determined for each perturbed parameter vector. The parameter vector is modified by moving it in the parameter vector space in a direction that maximizes a diversity metric. Other neural networks can be trained for new tasks or new environments using the parameter vector of the neural network.

Abstract: Various examples are directed to a service assignment system for providing transportation services. The service assignment system may receive a transportation service request from a user. The transportation service request may describe a transportation service having a service start location and a service end location. The service assignment system may generate a plurality of routes for executing the transportation service. The service assignment system may send proposed route data describing at least a portion of the plurality of routes to a first autonomous vehicle (AV). The service assignment system may receive, from the first AV, route data describing a first route to execute the transportation service and send, to the first AV, instruction data instructing the first AV to begin executing the transportation service.

Abstract: Various examples are directed to routing autonomous vehicles. A processor unit accesses first routing graph modification data and second routing graph modification data. The first routing graph modification data based at least in part on first vehicle capability data describing a first type of autonomous vehicle and the second routing graph modification data based at least in part on second vehicle capability data describing a second type of autonomous vehicle. The processor unit accesses routing graph data describing a plurality of graph elements and generates a first route for a first autonomous vehicle of the first type based at least in part on the first routing graph modification data and the routing graph data. The processor unit also generates a second route for a second autonomous vehicle of the second type based at least in part on the second routing graph modification data and the routing graph data.

Abstract: Systems and methods of configuring and using a machine-learned safety risk model to predict a corresponding risk of vehicular collision for different candidate routes are disclosed herein. In some example embodiments, a computer system obtains accident data and feature data for historical routes that have been communicated electronically as navigation guidance, trains a safety risk model using the accident data and the feature data of the historical routes as training data in a machine learning process, and then evaluates one or more routing algorithms by generating a corresponding set of routes for each routing algorithm and generating a corresponding performance measurement for each set of routes using the trained safety risk model.

Abstract: A system provides a set of data to a user application on a computing device of a user, the set of data being used by the mobile computing device to display a map indicating the current location of the mobile computing device on the map, an estimated arrival time for a vehicle to rendezvous with the user, and a user interface feature operable by the user to request transport. The system can select a driver to fulfill the user's transport request and prior to transport being provided to the user, the system tracks a current location of the selected driver as the selected driver progresses to the pickup location and provides progress information of the selected driver to the user. Upon the user being picked up by the selected driver, the system tracks the selected driver from the pickup location to a drop-off location.

Abstract: Systems and methods for dynamically re-ranking a pickup point are provided. The system receives a request for transportation service from a device of a rider that includes a selected pickup point. In response, the system establishes the transportation service including assigning a driver. Based on detecting that the transportation service is in a pre-pickup state, the system performs a re-rank analysis. The re-rank analysis determines whether an alternative pickup point exceeds the selected pickup point on an objective function by more than a predetermined threshold. The determining is based on a last known location of the rider, a last known location of the driver, and a predicted driver routeline. Responsive to the alternative pickup point exceeding the selected pickup point on the objective function by more than the predetermined threshold, the system causes presentation, on a device of the rider, of a suggestion to change to the alternative pickup point.

Abstract: Systems and methods herein describe a network system for generating inferred accurate locations. The systems and methods receive a transportation trip request from a first computing device that includes a target address, access a first plurality of historical location data and a second plurality of historical data, generate clustered location data using the first plurality of historical location data and the second plurality of historical location data, select a subset of cluster locations from the clustered location data, determine an inferred accurate location address, and modify the transportation trip request by associating the inferred accurate location address with the target address.

Abstract: Systems and methods for providing user control of alternate routes are provided. In example embodiments, a networked system receives a ride request from a user that indicates a drop-off location. The networked system identifies a current location of a user (e.g., a rider) and determines a plurality of routes from the current location of the user to a drop-off location. The plurality of routes is displayed on a user interface of a device of the user. In response, a selection of a route from the plurality of routes is received by the networked system. The networked system then causes presentation of a driving route corresponding to the selected route on a device of a driver and the device of the user.

Abstract: A system can receive a request for transport from a computing device of a user and select a service provider to provide transport for the user to a destination location. The system can receive location data from the computing device of the service provider. After the service provider picks up the user, the system can determine whether the service provider progresses toward the destination location in accordance with a set of progress conditions. Based at least in part on determining that the service provider has not progressed toward the destination location in accordance with the set of progress conditions, the system can adjust a fare for providing transport for the user to the destination location.

Abstract: Systems and methods for strategic routing and distribution of orders for multi-location merchants. The system can obtain data including a service request for multiple items and determine if the inventory at multiple store locations includes a first and second item. The method includes coordinating two couriers fulfilling the request. The method includes assigning a first courier to transport a first item from a first location to a second location and a second courier to transport a first item and second item from the second location to a destination location. The method includes determining a trigger time to send an assignment to a second courier device based on an estimated time for the second courier to travel to the second location and obtain the second item. The method includes automatically triggering transmission of the second assignment based on a progress of the first courier and the trigger time.