Method for selecting an optimized trajectory

Abstract

A method for generating a signal for transferring a partly or highly automated vehicle into a safe system state at a target site. First, a need to transfer the vehicle into a safe system state is ascertained. A vehicle state is then determined, the vehicle state encompassing the current vehicle position. At least one target site is ascertained. Travel trajectories are ascertained from the current vehicle position to the at least one target site. The travel trajectories are related. One of the travel trajectories is selected based on the rating that has been carried out. A signal is generated on the basis of the selected travel trajectory.

Claims

1. A method for generating a signal for transferring a partly or highly automated vehicle into a safe system state at a target site, the method comprising: ascertaining, via a processor, a need to transfer the vehicle into a safe system state; determining, via the processor, a vehicle state, the vehicle state including a current vehicle position; ascertaining, via the processor, at least one target site; ascertaining, via the processor, travel trajectories from the current vehicle position to the at least one target site; rating, via the processor, the travel trajectories; selecting, via the processor, one of the travel trajectories based on the rating; generating, via the processor, the signal based on the selected travel trajectory; and transferring, via the processor and based on the signal, a partly or highly automated vehicle into a safe system state at a target site; wherein expected localization accuracies along the ascertained travel trajectories are calculated, and those localization accuracies are used to rate the travel trajectories.

2. The method as recited in claim 1, wherein at least one of: i) a vehicle state, and ii) a driver state, are ascertained in order to ascertain the need.

3. The method as recited in claim 1, wherein a performance of the vehicle is ascertained, and the ascertained performance is used to rate the travel trajectories.

4. The method as recited in claim 1, wherein the expected localization accuracy along a travel trajectory is calculated on the basis of landmarks along that travel trajectory.

5. The method as recited in claim 1, wherein a functionality of environmental sensor equipment present in the vehicle is determined in the context of ascertaining the performance of the vehicle, and that functionality is used for calculation of the expected localization accuracy.

6. The method as recited in claim 1, wherein curves in the ascertained travel trajectories are determined, and the determined curves are used to rate the travel trajectories.

7. The method as recited in claim 1, wherein a first and at least one second target site are ascertained, the ascertained target sites being rated.

8. The method as recited in claim 7, wherein selection of one of the travel trajectories is accomplished based additionally on the rating of the at least two target sites.

9. The method as recited in claim 7, wherein a probability value that is calculated from a probability of at least one of: i) a collision with other traffic participants, and ii) interference with other traffic participants, at that target site is used for ascertaining the at least one target site.

10. The method as recited in claim 1, wherein map information is taken into consideration in ascertaining the at least one target site.

11. An apparatus for generating a signal for transferring a partly or highly automated vehicle into a safe system state at a target site, comprising: a transfer device having a processor configured to perform the following: ascertaining, via the processor, a need to transfer the vehicle into a safe system state; determining, via the processor, a vehicle state, the vehicle state including a current vehicle position; ascertaining, via the processor, at least one target site; ascertaining, via the processor, travel trajectories from the current vehicle position to the at least one target site; rating, via the processor, the travel trajectories; selecting, via the processor, one of the of the travel trajectories based on the rating; generating, via the processor, the signal based on the selected travel trajectory; and transferring, via the processor and based on the signal, a partly or highly automated vehicle into a safe system state at a target site; wherein expected localization accuracies along the ascertained travel trajectories are calculated, and those localization accuracies are used to rate the travel trajectories.

12. A non-transitory computer-readable storage medium on which is stored a computer program, which is executable by a processor, comprising: a program code arrangement having program code for generating a signal for transferring a partly or highly automated vehicle into a safe system state at a target site, by performing the following: ascertaining, via the processor, a need to transfer the vehicle into a safe system state; determining, via the processor, a vehicle state, the vehicle state including a current vehicle position; ascertaining, via the processor, at least one target site; ascertaining, via the processor, travel trajectories from the current vehicle position to the at least one target site; rating, via the processor, the travel trajectories; selecting, via the processor, one of the travel trajectories based on the rating; generating, via the processor, the signal based on the selected travel trajectory; and transferring, via the processor and based on the signal, a partly or highly automated vehicle into a safe system state at a target site; wherein expected localization accuracies along the ascertained travel trajectories are calculated, and those localization accuracies are used to rate the travel trajectories.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a schematic flow chart of an exemplifying method.

(2) FIG. 2 shows an example of the rating of different travel trajectories.

DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS

(3) FIG. 1 depicts an example of execution of the example method in a vehicle 201. The vehicle is equipped with a corresponding first apparatus on which a computer program is configured to execute the method. Vehicle 201 is furthermore equipped with sensors for environmental sensing, and optionally also with sensors for interior sensing. A second apparatus can also be provided in vehicle 201, which apparatus receives the signal, generated on the basis of a selected travel trajectory 204, 205, of the first apparatus and, based on that signal, carries out an application of control to corresponding actuators of vehicle 201 in order to guide the latter automatically along the selected trajectory 204, 205.

(4) The method starts in step 101.

(5) In step 102 the need to transfer vehicle 201 into a safe system state is ascertained. For this, the driver state and/or the vehicle state is ascertained. The interior sensing system of vehicle 201 is used to ascertain the driver state.

(6) Monitoring of the driver state, or of the state of a vehicle occupant, is accomplished here using known methods. The vehicle occupant is observed with an interior camera, and his or her state is evaluated on the basis of the camera images. The latter can supply, for example, eye movement, body temperature, body posture, unusual movement sequences, and further details. In addition, sensors can be provided in vehicle 201 which measure current values of the vehicle occupant that are useful for medical diagnosis, among them his or her blood pressure, blood sugar level, and body temperature. EKG data of the vehicle occupant can also be monitored.

(7) Important driving functions and sensors can be investigated with regard to their functionality in order to ascertain the vehicle state. For this, for example, error messages of individual vehicle subsystems can be detected as an input signal, and their state can be deduced from those signals. Logical associations are also conceivable, for example in which combining the signals of specific subsystems indicates faults in other subsystems.

(8) If it is established in the context of ascertaining the vehicle state that there exists a limitation of the system on the basis of which safety could be impaired, for example in the form of failure of an image sensor, then a need to transfer vehicle 201 into a safe system state is consequently identified. The same is true for the case in which it is established in the context of ascertaining the driver state that, for example because of health problems, the driver is no longer capable of taking control of the vehicle or that assistance is needed.

(9) In general, it is possible to ascertain on the basis of stored scenarios whether the currently existing state (driving state and driver state) generates such a need.

(10) If it is established in step 102 that a need to transfer vehicle 201 into a safe system state exists, then the driving state of the vehicle is determined in step 103. Here the current vehicle position, travel direction, and optionally the lane currently being traveled in, the current speed, and current traffic situation are ascertained. The traffic situation can be ascertained by collecting all information regarding those roadways in the close vicinity which could be traveled by vehicle 201. The information encompasses, for example, traffic jams, construction sites, and similar traffic impediments or vehicle accumulations.

(11) In step 104, target sites 208, 209 at which vehicle 201 can be transferred into a safe system state are ascertained. On the one hand it is possible to use for this the environmental sensor equipment, with which target sites 208, 209 located, for example, within visual range can be ascertained or a shoulder or parking space can be looked for. In addition, further target sites 208, 209 that can be arrived at can be ascertained with the aid of map information and the current vehicle position. The travel direction, lane currently being traveled in, and current speed can also be taken in consideration, inter alia, for this. On expressways in particular, the travel direction is essential, and depending on the lane being traveled in and on the speed, it may be that specific parking areas can no longer be reached because a lane change and a necessary deceleration of the vehicle are no longer possible. The current traffic situation can also be taken into consideration in the selection of target sites 208, 209, since it can provide information as to how much time is needed to convey the vehicle to a corresponding target site 208, 209.

(12) In this example, probability values that indicate the probability of a collision with other traffic participants at the corresponding target sites 208, 209 are used to assist in ascertaining target sites 208, 209. The calculation of this probability can be based on accident research data with which it is possible to determine for certain locations, based on accident data, how frequently accidents occur at a comparable location (for example a shoulder). The probability value does not need to be expressed as a percentage, but instead can contain, for example, three steps (low, medium, and high probability). An association between the numerical values from accident research and the aforementioned steps can be achieved, for example, using fuzzy logic.

(13) In step 105, travel trajectories 204, 205 from the current vehicle position to the ascertained target sites 208, 209 are ascertained. Depending on the target site 208, 209, the environmental sensor equipment and/or map material is used for this. A known approach, which is described for example in the publication of Fiorini et al. from the existing art set forth above, is used in this example to ascertain travel trajectories 204, 205. According to this approach, the states of the other vehicles can also be taken into consideration in determining trajectories 204, 205. From a possible set of trajectories 204, 205 to a target site 208, 209, i.e. a set that can be traveled to in consideration e.g. of the vehicle model and the available space, n trajectories are selected based on a search algorithm using a minimization method.

(14) In step 106 the performance of vehicle 201 is ascertained. As in the case of ascertaining the vehicle state, here the functionality of the environmental sensor equipment and of the components for automatic control of vehicle 201 is checked. Ascertainment is effected especially with a view toward detecting limitations of vehicle 201 in a context of automatic control. Important information here includes, for example, information regarding failed sensors that limit the viewing region, and limited capabilities for controlling the vehicle, for example steering, braking, or speed.

(15) In step 107 the ascertained travel trajectories 204, 205 are rated. For this, the expected localization accuracy of vehicle 201 along travel trajectories 204, 205 is calculated. This can be done, for example, on the basis of landmarks 206, 207 disposed along travel trajectory 204, 205 which are used to localize vehicle 201. A probabilistic model offers the basis for estimating the expected localization accuracy along travel trajectory 204, 205. For derivation of the model it is assumed that the global vehicle position is obtained from matching between landmarks 206, 207 of a global localization map and corresponding landmark measurements. Error propagation can be carried out for the algorithms to be used for this. The resulting probabilistic model describes the correlation between the input variables (corresponding landmarks, error model for landmark measurement) and the expected uncertainty (variance) in the estimate of the global vehicle position. The model is provided in closed analytical form (i.e. iterative calculation methods are not required) and provides estimates of the localization accuracy for a given assemblage of landmarks 206, 207 which are known from map information (in this case from a highly accurate map). Thanks to the analytically closed form, rating of the trajectories is associated with a low calculation outlay.

(16) The performance of vehicle 201 can furthermore be included in the rating of travel trajectories 204, 205. For example, if optical parts of an installed camera are dirty or if a camera for observation of a specific region fails, landmarks 206, 207 that can no longer be detected because of the limitation of the sensor equipment then will not be taken into consideration in the calculation of localization accuracy. In addition, possible concealment effects of landmarks 206, 207 due to static objects along a specific travel trajectory 204, 205 can also be included in the calculation, with the result that, for example, a travel trajectory 204, 205 in a different lane would be preferred.

(17) Concealment effects due to moving objects can also be incorporated into the rating of travel trajectories 204, 205. For example, the probability of concealment of objects on the right side of a two-lane roadway is to be estimated to be higher if vehicle 201 is traveling in the left lane than if it were traveling in the right lane. Motor vehicles 203, 203 and trucks 202, 203 in the right lane can conceal specific landmarks 206, 207.

(18) If further sensors are limited in terms of their functionality, for example GPS sensors, sensors for location via mobile radio signals, yaw rate sensors, rotation rate sensors, acceleration sensors, wheel rotation speed sensors, radar, lidar, or ultrasound sensors, or any other sensors usable for determining a driving state, they can have corresponding effects on the rating of travel trajectories 204, 205.

(19) If vehicle 201 is limited in terms of its freedom of movement, for example if only certain steering angles can be achieved, this can likewise be included in the rating of travel trajectories 204, 205. In this case a travel trajectory 204, 205 having too sharp a curve would be excluded.

(20) In addition to the localization of vehicle 201 with the aid of landmarks 206, 207 that are detected by the environmental sensor equipment, odometry in vehicle 201 can also be used as an alternative localization method. Because sharp curves in travel trajectory 204, 205 can have a negative effect on the accuracy of this localization method, when odometry is used, travel trajectories 204, 205 can be rated on the basis of curves that occur.

(21) Lastly, before a travel trajectory 204, 205 is selected, the ratings for target sites 208, 209 corresponding to travel trajectories 204, 205 can be included in the rating of travel trajectories 204, 205. For example, if two travel trajectories 204, 205 to different target sites 208, 209 have similarly good ratings, the location of target site 208, 209 can be definitive for the final selection of trajectory 204, 205. The localization accuracy of target site 208, 209, and the probability of a collision with other traffic participants or of interference, can enter into the rating of target site 208, 209.

(22) In step 108, a travel trajectory 204, 205 that is to be traveled to is selected. The selection is made based on the rating arrived in step 107, the best-rated trajectory 204, 205 being selected.

(23) In step 109 a signal is generated on the basis of the travel trajectory selection carried out in step 108. The signal can contain information regarding waypoints, speeds, and accelerations. These indications can be provided in the vehicle reference system or in a map reference system. In this exemplifying embodiment two forms are preferably used: 1) If stable communication with a map server and a sufficiently functional environmental sensing system can be ensured, a travel trajectory 204, 205 in the map reference system is then used. Map-relative localization can thus be carried out. As a result, there is no drift in the estimate of the vehicle position due to inaccurate odometry. 2) The preferred variant form for a vehicle system having considerable functional limitations (particularly in terms of communication with a map server) and limited environmental awareness is an indication of the vehicle trajectory in the vehicle reference system, since safe travel is thereby possible even without map material.

(24) Depending on the embodiment of the method according to the present invention and how it is used, the signal can be sent directly to a corresponding vehicle controller that applies control to corresponding actuators of vehicle 201 so that the latter is steered along travel trajectory 204, 205.

(25) Alternatively, the signal can also be sent to a further control device that processes the information and then initiates automatic travel along the travel trajectory.

(26) The method ends at step 110.

(27) The sequence of steps in this method is also modifiable. For example, performance can already be ascertained before travel trajectories 204, 205 are ascertained, and thus already included in travel trajectory planning.

(28) FIG. 2 depicts by way of example the criteria on the basis of which the travel trajectory can be selected.

(29) In this example, vehicle 201 is traveling on a three-lane expressway and thus has the option of driving in all three lanes. In the left lane a truck 202 is approaching from behind and limiting the environmental sensor equipment of vehicle 201. The same applies to truck 203 on the right side of vehicle 201, as a result of which the environmental sensor equipment of vehicle 201 cannot detect landmark 206.

(30) FIG. 2 depicts two target sites 208, 209 and a travel trajectory 204, 205 respectively corresponding to target sites 208, 209. Travel trajectories 204, 205 are rated on the basis of the expected localization accuracy of vehicle 201 along trajectories 204, 205, calculation of the localization accuracy being accomplished on the basis of landmarks 206, 207 that are present.

(31) Because the rating of trajectory 204 ends up being higher because of the plurality of detectable landmarks, this travel trajectory 204 is selected and a corresponding signal is generated on the basis of that selection.