IPC분류정보
국가/구분 |
United States(US) Patent
등록
|
국제특허분류(IPC7판) |
|
출원번호 |
US-0558573
(2009-09-13)
|
등록번호 |
US-8463461
(2013-06-11)
|
발명자
/ 주소 |
- Estkowski, Regina I.
- Wilson, Jr., Robert C.
- Whitley, Ted D.
|
출원인 / 주소 |
|
대리인 / 주소 |
|
인용정보 |
피인용 횟수 :
1 인용 특허 :
4 |
초록
▼
According to an embodiment herein, a method of predicting a trajectory of an aerospace vehicle comprises accessing an observation of a state of the vehicle from sensor data; and using a computing system to predict different possible future positions and attitudes of the vehicle, including using the
According to an embodiment herein, a method of predicting a trajectory of an aerospace vehicle comprises accessing an observation of a state of the vehicle from sensor data; and using a computing system to predict different possible future positions and attitudes of the vehicle, including using the sensor data and associated latencies to determine a set of vehicle state transitions. Each state transition in the set is computed as a function of estimated latency. The method further comprises using the computing system to update a prior distribution of the vehicle state with the state transitions. Consequently, the predicted trajectory is compensated for latency.
대표청구항
▼
1. A method of predicting a trajectory of an aerospace vehicle, the method comprising: accessing an observation of a state of the vehicle from sensor data;using a computing system to predict different possible future positions and attitudes of the vehicle, including using the sensor data and associa
1. A method of predicting a trajectory of an aerospace vehicle, the method comprising: accessing an observation of a state of the vehicle from sensor data;using a computing system to predict different possible future positions and attitudes of the vehicle, including using the sensor data and associated latencies to determine a set of vehicle state transitions, each state transition in the set computed as a function of estimated latency; andusing the computing system to update a prior distribution of the vehicle state with the state transitions;whereby the predicted trajectory is compensated for latency. 2. The method of claim 1, wherein a particle filter is used to predict the possible future positions and attitudes of the vehicle; and wherein each state transition particle is a function of latency. 3. The method of claim 2, wherein each state transition particle is a function of a vehicle state delta and a weight; wherein the delta occurs over a period matching the latency, and wherein each weight is based on a probability of latency. 4. The method of claim 1, wherein accessing the observations includes accessing the sensor data from multiple data links having different latency characteristics. 5. The method of claim 4, wherein latencies and uncertainties associated with each data link are estimated and used to determine the state transitions. 6. The method of claim 1, wherein predicting the future positions and attitudes includes determining a set of state transitions from the sensor data and latency probabilities, and performing particle filtering on the distribution to determine weights for the state transitions (particles) in the distribution. 7. The method of claim 6, wherein a plurality of number of state transitions are pre-computed and accessed at run-time according to observed indicators and data latencies; and wherein the pre-computed state transitions are used to obtain the distribution. 8. The method of claim 7, wherein at least some variables to the particle filter are marginalized. 9. The method of claim 1, wherein using the computing system includes: accessing a current (Nth) posterior particle distribution for a vehicle state;computing a set of state transition particles with probability distribution, where each state transition particle is a function of estimated latency;operating on the current vehicle state distribution with the state transition set to obtain a prior distribution; andcomputing particle weights to obtain an N+1th posterior particle distribution. 10. The method of claim 9, wherein the state transitions use a time step that is a function of latency, and wherein particle weights are a function of a probability of latency. 11. The method of claim 10, wherein the time step of the state transitions is chosen to match at least a sum of a link latency and output latency. 12. The method of claim 10, wherein the set of particles in the current posterior distribution is reduced in order to increase computational speed. 13. A system for predicting movement of a vehicle in airspace, comprising a computer programmed to: receive multiple observations over multiple data links about vehicle states, including estimated latencies associated with the observations; andupdate a vehicle state particle distribution with a set of state transition particles, each state transition particle being a function of latency associated with the observations, where the spread of the updated distribution reflects uncertainties due to the latencies of the data that is used in generating the distribution. 14. The system of claim 13, wherein a particle filter is used to update the vehicle state and wherein each state transition particle is a function of latency. 15. The system of claim 14, wherein each state transition particle is a function of a vehicle state delta and a weight; wherein the delta occurs over a period matching the latency, and wherein each weight is based on a probability of latency. 16. The system of claim 13, wherein updating the vehicle distribution includes: accessing a current (Nth) posterior particle distribution for a vehicle state;computing a set of state transition particles with probability distribution, where each state transition particle is a function of estimated latency;operating on the current vehicle state distribution with the state transition set to obtain a prior distribution; andcomputing particle weights to obtain an N+1th posterior particle distribution. 17. An article comprising computer memory encoded with data for causing a computer system to predict different possible future positions and attitudes of an aerospace vehicle, including using the sensor data and associated latencies to determine a set of vehicle state transitions, each state transition in the set computed as a function of estimated latency; and updating a prior distribution of the vehicle state with the state transitions, whereby the predicted trajectory is compensated for latency. 18. The article of claim 17, wherein a particle filter is used to predict the possible future positions and attitudes of the vehicle; and wherein each state transition particle is a function of latency. 19. The article of claim 18, wherein each state transition particle is a function of a vehicle state delta and a weight; wherein the delta occurs over a period matching the latency, and wherein each weight is based on a probability of latency. 20. The article of claim 17, wherein updating the vehicle distribution includes: accessing a current (Nth) posterior particle distribution for a vehicle state;computing a set of state transition particles with probability distribution, where each state transition particle is a function of estimated latency;operating on the current vehicle state distribution with the state transition set to obtain a prior distribution; andcomputing particle weights to obtain an N+1th posterior particle distribution.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.