May Mobility has been granted a patent for a system that allows autonomous vehicles to operate with incomplete environmental information. The system involves receiving inputs, identifying known objects and blind regions in the vehicle’s environment, inserting virtual objects into the blind regions, and operating the vehicle based on these virtual objects. The patent also includes a method for updating the environmental representation of the vehicle based on sensor data and operating the vehicle accordingly. GlobalData’s report on May Mobility gives a 360-degree view of the company including its patenting strategy. Buy the report here.
According to GlobalData’s company profile on May Mobility, AI assisted drone control was a key innovation area identified from patents. May Mobility's grant share as of September 2023 was 45%. Grant share is based on the ratio of number of grants to total number of patents.
Operating an autonomous vehicle with incomplete environmental information
A recently granted patent (Publication Number: US11745764B2) describes a method for updating the environmental representation of an autonomous vehicle based on vehicle sensor data. The method involves determining an environmental representation for the vehicle's environment, which includes a set of known objects. The method then identifies unknown regions within the environmental representation and determines their intersection with conflict zones labeled within a map. Based on this intersection, the environmental representation is updated to include a set of virtual objects. The autonomous vehicle operates based on this updated environmental representation.
The method also includes determining the labeled conflict zones based on the behavior associated with the autonomous vehicle. A simulation is performed using the known objects and virtual objects, and the autonomous vehicle is controlled based on the simulation. The simulation can include a forward simulation performed at a predetermined cycle frequency.
To update the environmental representation, the method identifies a subset of unknown regions that overlap with the labeled conflict zones and includes a virtual object within each unknown region of the subset. The selection of virtual objects is based on the size parameter of each unknown region.
The method assigns parameters to the virtual objects, which are determined based on parameters associated with the known objects. These parameters can include speed, which is determined based on information associated with the known objects, such as historical speeds.
The environmental representation characterizes the environment of the autonomous vehicle, and the update of the representation to include virtual objects is based on the probability associated with each unknown region. The update occurs when the probability exceeds a threshold, which can be determined based on historical sensor data associated with the location or known objects in the environment. The probability can also be based on a change in the number of known objects in the representation.
The identification of unknown regions is based on height metrics, where each unknown region has a height exceeding a predetermined threshold relative to the ground level of the environment.
Overall, this patent presents a method for updating the environmental representation of an autonomous vehicle using sensor data and virtual objects, allowing for more accurate and informed decision-making by the vehicle.