Method, electronic device, and autonomous driving system for error event analysis
A method for error event analysis performed by an electronic device includes: in response to an error event of an autonomous driving system, feeding a sequence of signal states, collected from the autonomous driving system into a prediction model, where the prediction model is trained based on signal states collected from a plurality of autonomous driving systems for predicting, based on a first signal state, a second signal state, the first signal state representing signals collected at a first time, and the second signal state representing signals collected at a second time succeeding the first time; converting the error event into an event signal state having signal values at an event time; comparing the second signal state with the event signal state; and in response to the second signal state matching the event signal state, tracing the error event back to the first signal state as a root cause.
1 . A method for error event analysis, performed by an electronic device, the method comprising:
in response to an error event of an autonomous driving system, feeding a sequence of signal states of the error event and collected from the autonomous driving system into a prediction model that has been trained, wherein the prediction model is used for predicting, based on a first signal state, a second signal state, the first signal state representing signals collected at a first time, and the second signal state representing signals collected at a second time succeeding the first time;
converting the error event into an event signal state having signal values at an event time;
comparing the second signal state with the event signal state;
in response to the second signal state matching the event signal state, tracing the error event back to the first signal state as a root cause, by performing:
searching the sequence of signal states of the error event for a pattern that includes the first signal state and the second signal state,
determining that the matched event signal state appears in the pattern,
using the matched event signal state as the second signal state, wherein the second signal state is predicted from the first signal state, and
identifying the first signal state in the pattern corresponding to the second signal state as the root cause for the error event;
after identifying the first signal state as the root cause for the error event in a first tracing iteration, identifying another first signal state as another root cause for the first signal state identified in the first tracing iteration;
repeatedly identifying additional first signal state to obtain a chain of first signal states, the chain of signal states being considered as the root cause for the error event;
storing a plurality of signal states predicted by the prediction model such that the error event can be traced back to both an immediate signal state that causes the prediction model to predict the error event and a past signal state that causes the prediction model to predict the immediate signal state; and
controlling the autonomous driving system by performing an action to prevent the error event from reoccurring based on the root cause.
2 . The method according to claim 1 , further comprising:
outputting the root cause on a display of the electronic device.
3 . The method according to claim 1 , wherein:
the signal states include measurement data measured by sensors of the autonomous driving system and are collected at a predetermined time interval from the autonomous driving system to become the sequence of signal states that are used in error event analysis.
4 . The method according to claim 3 , wherein:
the sequence of signal states further includes abstraction signals that are extracted from the measurement data measured by sensors of the autonomous driving system and local knowledges including local maps and local rules and regulations.
5 . The method according to claim 1 , wherein:
matching the second signal state with the event signal state includes matching signal values of the second signal state with signal values of the event signal state.
6 . The method according to claim 5 , wherein:
the event signal state includes signal values only for error event related signals; and
matching the signal values of the second signal state with the signal values of the event signal state includes matching the signal values of a portion of signals in the second signal state that correspond to the error event related signals with the signal values of the error event related signals in the event signal state.
7 . The method according to claim 1 , wherein the pattern including the first signal state and the second signal state is previously learned by the prediction model.
8 . The method according to claim 1 , wherein:
the prediction model is a Hierarchical Temporal Memory (HTM) model; and
each signal state is a sparsely distributed matrix of signal values.
9 . An electronic device for error event analysis, comprising a processor and a memory coupled to the processor and storing program instructions, wherein when being executed by the processor, the program instructions cause the processor to:
in response to an error event of an autonomous driving system, feed a sequence of signal states of the error event and collected from the autonomous driving system into a prediction model that has been trained, wherein the prediction model is used for predicting, based on a first signal state, a second signal state, the first signal state representing signals collected at a first time, and the second signal state representing signals collected at a second time succeeding the first time;
convert the error event into an event signal state having signal values at an event time;
compare the second signal state with the event signal state;
in response to the second signal state matching the event signal state, trace the error event back to the first signal state as a root cause, by performing: searching the sequence of signal states of the error event for a pattern that includes the first signal state and the second signal state, determining that the matched event signal state appears in the pattern, using the matched event signal state as the second signal state, wherein the second signal state is predicted from the first signal state, and identifying the first signal state in the pattern corresponding to the second signal state as the root cause for the error event;
after identifying the first signal state as the root cause for the error event in a first tracing iteration, identify another first signal state as another root cause for the first signal state identified in the first tracing iteration;
repeatedly identify additional first signal state to obtain a chain of first signal states, the chain of signal states being considered as the root cause for the error event;
store a plurality of signal states predicted by the prediction model such that the error event can be traced back to both an immediate signal state that causes the prediction model to predict the error event and a past signal state that causes the prediction model to predict the immediate signal state; and
control the autonomous driving system to perform an action to prevent the error event from reoccurring based on the root cause.
10 . The electronic device according to claim 9 , wherein the processor is further configured to:
output the root cause on a display of the electronic device.
11 . The electronic device according to claim 9 , wherein:
the signal states include measurement data measured by sensors of the autonomous driving system and are collected at a predetermined time interval from the autonomous driving system to become the sequence of signal states that are used in error event analysis.
12 . The electronic device according to claim 11 , wherein:
the sequence of signal states further includes abstraction signals that are extracted from the measurement data measured by sensors of the autonomous driving system and local knowledges including local maps and local rules and regulations.
13 . The electronic device according to claim 9 , wherein:
matching the second signal state with the event signal state includes matching signal values of the second signal state with signal values of the event signal state.
14 . The electronic device according to claim 13 , wherein:
the event signal state includes signal values only for error event related signals; and
matching the signal values of the second signal state with the signal values of the event signal state includes matching the signal values of a portion of signals in the second signal state that correspond to the error event related signals with the signal values of the error event related signals in the event signal state.
15 . The electronic device according to claim 9 , wherein the pattern including the first signal state and the second signal state that is previously learned by the prediction model.
16 . An autonomous driving system, comprising an electronic device for error event analysis, wherein:
the electronic device includes a processor and a memory coupled to the processor and storing program instructions, wherein when being executed by the processor, the program instructions cause the processor to:
in response to an error event of an autonomous driving system, feed a sequence of signal states of the error event and collected from the autonomous driving system into a prediction model that has been trained, wherein the prediction model is used for predicting, based on a first signal state, a second signal state, the first signal state representing signals collected at a first time, and the second signal state representing signals collected at a second time succeeding the first time;
convert the error event into an event signal state having signal values at an event time;
compare the second signal state with the event signal state;
in response to the second signal state matching the event signal state, trace the error event back to the first signal state as a root cause, by performing: searching the sequence of signal states of the error event for a pattern that includes the first signal state and the second signal state, determining that the matched event signal state appears in the pattern, using the matched event signal state as the second signal state, wherein the second signal state is predicted from the first signal state, and identifying the first signal state in the pattern corresponding to the second signal state as the root cause for the error event;
after identifying the first signal state as the root cause for the error event in a first tracing iteration, identify another first signal state as another root cause for the first signal state identified in the first tracing iteration;
repeatedly identify additional first signal state to obtain a chain of first signal states, the chain of signal states being considered as the root cause for the error event;
store a plurality of signal states predicted by the prediction model such that the error event can be traced back to both an immediate signal state that causes the prediction model to predict the error event and a past signal state that causes the prediction model to predict the immediate signal state; and
control the autonomous driving system to perform an action to prevent the error event from reoccurring based on the root cause.
17 . The system according to claim 16 , wherein:
the signal states include measurement data measured by sensors of the autonomous driving system and are collected at a predetermined time interval from the autonomous driving system to become the sequence of signal states that are used in error event analysis.
18 . The system according to claim 17 , wherein:
the sequence of signal states further includes abstraction signals that are extracted from the measurement data measured by sensors of the autonomous driving system and local knowledges including local maps and local rules and regulations.