Method for recognizing objects in an environment of a vehicle
A method recognizes objects in an environment of a vehicle including a sensor system having at least one sensor unit for registering the environment and an evaluation unit for evaluating sensor data provided by the at least one sensor unit. The method includes receiving the sensor data in the evaluation unit, the sensor data including a plurality of chronologically successive measurements; inputting the sensor data into a machine learning module; and outputting an object state of at least one object estimated based on the sensor data by way of the machine learning module. The method further includes determining a plurality of chronologically successive future object states based on the estimated object state; ascertaining deviations between the future object states and measurements chronologically corresponding to the future object states based on the sensor data; and correcting a machine learning algorithm of the module based on the deviations.
1. A method for recognizing objects in an environment of a vehicle, the vehicle comprising a sensor system having at least one sensor unit for registering the environment and an evaluation unit for evaluating sensor data provided by the at least one sensor unit, the method comprising:
receiving the sensor data in the evaluation unit, the sensor data comprising a plurality of chronologically successive measurements generated by the at least sensor unit;
inputting the sensor data into a machine learning algorithm of a machine learning module;
outputting an estimated object state of at least one object estimated based on the sensor data by way of the machine learning module;
determining a plurality of chronologically successive future object states based on the estimated object state;
transforming the future object states into a coordinate system of the at least one sensor unit to obtain a transformed future object state for each of the future object states;
ascertaining deviations between the transformed future object states and measurements chronologically corresponding to the transformed future object states based on the sensor data; and
correcting the machine learning algorithm based on the deviations.
2. The method according to claim 1 , wherein:
the sensor data comprise measurements of at least two different sensor units of the sensor system;
the future object states are transformed into coordinate systems of the at least two different sensor units; and
the transformed future object state and the measurement between which the deviation is to be ascertained correspond in their coordinate systems.
3. The method according to claim 2 , wherein each of the future object states is transformed into a different coordinate system.
4. The method according to claim 1 , wherein:
the sensor data comprise a first series of measurements and a second series of measurements chronologically following the first series;
the estimated object state describes a state of the object at an estimation point in time;
each of the future object states describes a state of the object at a future point in time following the estimation point in time; and
the measurements of the second series are each associated with one of the future points in time.
5. The method according to claim 4 , wherein the estimation point in time is before a point in time of a first measurement of the second series.
6. The method according to claim 4 , wherein the estimation point in time corresponds to a point in time of a last measurement of the first series.
7. The method according to claim 1 , further comprising:
determining weighted deviations from the deviations, the weighted deviation based on a respective time interval between the future object states and the estimated object state,
wherein the weighted deviations are input into the machine learning algorithm of the machine learning module.
8. The method according to claim 7 , wherein:
a sum is calculated from the weighted deviations; and
the sum is input into the machine learning algorithm of the machine learning module.
9. The method according to claim 1 , wherein the evaluation unit is configured to execute the method.
10. A vehicle system, including:
a sensor system having at least one sensor unit configured to register an environment of a vehicle; and
an evaluation unit operably connected to the sensor system, and configured to evaluate sensor data received from the at least one sensor unit, the sensor data comprising a plurality of chronologically successive measurements;
wherein the evaluation unit is configured to recognize objects in an environment of the vehicle and is configured to:
input the sensor data into a machine learning algorithm of a machine learning module;
output an estimated object state of at least one object estimated based on the sensor data by way of the machine learning module;
determine a plurality of chronologically successive future object states based on the estimated object state;
transform the future object states into a coordinate system of the at least one sensor unit to obtain a transformed future object state for each of the future object states;
ascertain deviations between the transformed future object states and measurements chronologically corresponding to the transformed future object states based on the sensor data; and
correct the machine learning algorithm based on the deviations.
11. The vehicle system according to claim 10 , wherein:
the sensor data comprise measurements of at least two different sensor units of the sensor system,
the evaluation unit is further configured to transform the future object states into coordinate systems of the at least two different sensor unit, and
the transformed future object state and the measurement between which the deviation is to be ascertained correspond in their coordinate systems.