Steering wheel hands-off detection method and system, and vehicle

Abstract

The present invention belongs to the technical field of intelligent vehicle interaction and relates to a steering wheel hands-off detection method and system, and a vehicle. According to the solution, a detection torque of a steering wheel for hands-off detection is obtained during intelligent driving, and the detection torque is compared with a calibration torque to obtain a steering wheel hands-off detection result, wherein the calibration torque is obtained by means of collecting steering wheel torques under the condition that the hands of a driver have been off the steering wheel during the intelligent driving process of a vehicle and performing statistical processing on the steering wheel torques. Since mechanical friction forces of steering systems of different vehicles are different, when torques generated by the mechanical friction forces are superimposed on a steering wheel torque sensor, it is difficult for the same set of software algorithms to be adapted to different vehicles, and thus after mass production, the recognition accuracy of different vehicles is uneven. The present application solves such problem, and improves the accuracy of hands-off detection.

Description

Steering wheel hands-off detection method, system and vehicle Technical Field

[0001] The present invention relates to a steering wheel hands-off detection method, system and vehicle, belonging to the technical field of intelligent vehicle interaction; in particular, to a torque-based steering wheel hands-off detection method for assisted driving. Background Art

[0002] Currently, most intelligent driving belongs to assisted driving below the L3 level. To ensure safety, drivers are required to keep their hands on the steering wheel when ADAS (Advanced Driver Assistance System) intervenes, so that they can take over the vehicle at any time in an emergency.

[0003] To address this, most ADAS systems are equipped with HOD (Hands Off Detection), which monitors whether the driver is holding the steering wheel. If the driver's hands leave the steering wheel, an alarm will be generated. If the driver remains hands-off for an extended period, the intelligent driving function will be disabled. While HOD is not currently mandatory, it is likely to be incorporated into national standards in the future.

[0004] Currently, the industry's main methods for detecting when a driver's hands are off the steering wheel include image-based visual recognition, capacitive touch sensors installed on the steering wheel, and steering wheel torque recognition. The first two methods require the installation of additional dedicated sensors, which increases the cost of the entire vehicle. In addition, the performance of machine vision recognition methods is easily affected by factors such as lighting and installation angle. The method based on steering wheel torque recognition can collect the torque sensor signal of the vehicle's electric power steering unit without increasing hardware costs. However, it is affected by the following two factors, resulting in low recognition accuracy:

[0005] ① When the driver grips the steering wheel lightly, the torque signal on the steering wheel changes less than when the hands are off. This is affected by hardware precision and is difficult to distinguish. This leads to low judgment accuracy and frequent switching of judgment states, affecting the assisted driving experience.

[0006] ② Due to mechanical friction during steering system rotation, especially in commercial vehicles, where mechanical assembly errors are large and operating environments and conditions are more severe than those of passenger cars, the magnitude of steering friction varies significantly between different vehicles (even those of the same model). When the torque generated by friction is superimposed on the torque sensor, the torque values ​​when the driver's hands are off the steering wheel vary significantly between different vehicles. This makes the same software algorithm difficult to adapt to different vehicles, resulting in varying recognition accuracy across different vehicles after mass production.

[0007] Chinese invention patent application publication number CN115123296A discloses a method for detecting hands-off steering wheels. This method utilizes a trained deep learning model to determine detected steering wheel torque signals, thereby improving the accuracy of hands-off steering wheel detection. However, this solution relies on an artificial intelligence algorithm, which requires extensive training data and imposes high hardware requirements for algorithm execution, hindering cost control.

[0008] A Chinese invention patent application, published under the publication number CN116161045A, discloses a method for detecting hands-off steering. This method first generates a hands-off status signal based on the system return torque value and the actual return torque value, essentially performing a preliminary check to determine whether the steering wheel has been released. This hands-off status signal is then combined with the requested and current steering wheel angles, as well as the actual return torque value, to determine whether the steering wheel is in a hands-off state, thereby improving the accuracy of determining whether the steering wheel has been released. This method determines hands-off by determining that the difference between the system return torque and the actual return torque exceeds a set value a certain number of times and for a preset time. This process is cumbersome and relies on numerous calibrated and real-time measured physical quantities, which is not conducive to improving recognition accuracy.

[0009] Summary of the Invention

[0010] The purpose of the present invention is to provide a method, system and vehicle for detecting hands-off steering wheel, so as to solve the problem of complex hands-off detection judgment process in the prior art.

[0011] To achieve the above object, the solution of the present invention includes:

[0012] A technical solution for a method for detecting hands-off steering wheels of the present invention obtains a detection torque of the steering wheel for hands-off detection during intelligent driving, and compares the detection torque with a calibration torque to obtain a steering wheel hands-off detection result; the calibration torque is obtained by collecting the steering wheel torque and performing statistical processing when the driver's hands are off the steering wheel during intelligent driving of the vehicle.

[0013] Furthermore, the comparison of the detected torque and the calibrated torque includes subtracting the calibrated torque from the detected torque to obtain the judgment torque, and comparing the judgment torque with one or more pre-set thresholds. At least when the judgment torque is less than a threshold, the steering wheel hands-off detection result is judged to be hands-off; at least when the judgment torque is greater than a threshold, the steering wheel hands-off detection result is judged to be not hands-off.

[0014] Furthermore, when the judgment torque is less than a threshold value and reaches a set time, the steering wheel hands-off detection result is judged to be hands-off.

[0015] Furthermore, within the set time, if it is determined that the torque is again greater than a set threshold, the steering wheel hands-off detection result is still the current hands-on state.

[0016] Furthermore, when the torque is judged to be greater than a threshold value and a set time is reached, the steering wheel hands-off detection result is judged to be not-off-hand.

[0017] Furthermore, within the set time, if it is determined that the torque is again less than a set threshold, the steering wheel hands-off detection result is still the current hands-off.

[0018] Furthermore, the calibration process for obtaining the calibrated torque is performed before the vehicle rolls off the production line or leaves the factory; or after the vehicle has been running for a period of time.

[0019] Furthermore, the statistical processing includes averaging the absolute values ​​of the steering wheel torque collected within a certain period of time to obtain the calibrated torque.

[0020] Furthermore, during the calibration process of obtaining the calibrated torque, the intelligent driving process of the vehicle reaches a set time.

[0021] A technical solution of a steering wheel hands-off detection system of the present invention includes a processor, which executes the steering wheel hands-off detection method described above.

[0022] A technical solution for a vehicle of the present invention includes the above-mentioned hands-off steering wheel detection system.

[0023] The beneficial effects of the present invention are:

[0024] To address the low torque detection recognition rate issue mentioned in the background, this invention builds upon existing technologies by implementing a torque self-learning process, either pre-shipment or periodically. During intelligent driving, the vehicle collects steering wheel torque when the driver's hands are off the steering wheel. This torque is then processed as a calibration torque to compensate for actual detection values ​​during hands-off detection. By employing torque self-learning compensation, this invention eliminates the influence of steering system friction torque on hands-off state recognition caused by individual vehicle differences, thereby improving hands-off detection accuracy.

[0025] At the same time, compared with the existing technical solutions listed in the background technology, the present invention is simple and reliable to implement, has low requirements on computing power, does not require new hardware, and has almost no impact on costs.

[0026] At the same time, when using the detection torque and calibration torque to make a hands-off judgment, the present invention further adopts a multi-state machine jump confirmation method, which further improves the detection accuracy and avoids frequent state jumps leading to frequent reminders or automatic exit of intelligent driving, affecting the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] FIG1 is a flow chart of a method for detecting hands-off steering wheel in Example 1 of the present invention;

[0028] FIG2 is a schematic diagram of the hands-off detection and judgment logic in Example 1 of the present invention. DETAILED DESCRIPTION

[0029] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0030] The concept of the present invention is to enable the vehicle to enter an intelligent driving mode while ensuring safety, and to drive the vehicle for a period of time without the driver holding the steering wheel. The intelligent driving system (such as an ADAS system) used to implement the intelligent driving function collects the steering wheel torque value during this period and performs statistical data processing to obtain a system friction torque that can reflect the friction condition of the vehicle's steering system. Due to changes in the steering system friction condition caused by assembly errors and the operating environment, the system friction torque of the steering systems of different vehicles is also different.

[0031] When the vehicle is in intelligent driving mode during normal driving, the intelligent driving system uses the system friction torque as a benchmark and compares it with the real-time monitored steering wheel torque value to make a judgment. This can eliminate the problem of low torque-based steering wheel hands-off detection accuracy caused by different steering system friction torques due to individual vehicle errors.

[0032] Moreover, the calibration and detection method of the present invention has simple logic and high detection accuracy, which avoids the problem of frequent alarms or repeated exits of the intelligent driving system due to detection errors, and greatly improves the experience of intelligent driving and human-vehicle interaction without increasing hardware costs.

[0033] Method Example 1:

[0034] A method for detecting hands-off steering wheel according to this embodiment is shown in FIG1 , and includes the following steps:

[0035] 1. System online calibration.

[0036] After a vehicle rolls off the production line, the intelligent driving system's hands-off steering detection module undergoes dynamic calibration. First, after triggering the calibration state, the driver enters intelligent driving mode. While safely driving in a dedicated area, on a closed road, or on an open road, the driver takes their hands off the steering wheel and uses the intelligent driving function for a specified time (t). The hands-off steering detection module collects the steering wheel torque during this time and calculates the average absolute value of the torque during this time (t) as the calibration torque value, TQ1. This torque value (TQ1) primarily represents the friction torque of the vehicle's steering system after leaving the factory. Due to assembly errors, the friction torque of the steering system varies from vehicle to vehicle. The hands-off steering detection module then writes the calibration torque value, TQ1, to non-volatile memory for subsequent system access.

[0037] The intelligent driving modes / functions referred to in this embodiment may include, for example, autonomous driving, where the vehicle's intelligent driving system fully controls the vehicle's control under various road and traffic conditions, or assisted driving, where the driver does not need to operate the steering wheel, such as adaptive cruise control or lane keeping under specific road conditions, or automated parking. In short, these modes / functions are situations where the driver does not need to operate the steering wheel, but must keep his hands on the steering wheel to ensure safety.

[0038] The corresponding intelligent driving system for realizing the intelligent driving function can be an ADAS system, or a hardware system that realizes the above-mentioned single function, such as an adaptive cruise control system, a lane keeping system, or an automatic parking system; it can also be integrated into the vehicle controller or other vehicle controllers, and the corresponding hardware system is the intelligent driving system in this embodiment.

[0039] The above steps are the system's online learning and calibration process. The subsequent steps are to enter the actual driving process. When the intelligent driving system takes over the vehicle, the steering wheel hands-off detection module performs hands-off detection based on the stored calibration torque value TQ1 that can reflect the friction torque of the vehicle steering system, and then performs actual driver hands-off detection.

[0040] 2. Driver hands-off detection signal collection and processing.

[0041] During actual vehicle operation, when the vehicle is in intelligent driving mode, the intelligent driving system takes over the vehicle, and at the same time, the steering wheel hands-off detection module reads the steering wheel torque signal in real time through the steering wheel motor assist system, filters the torque signal, and removes abnormal values ​​and noise. The torque absolute value corresponding to the torque signal is obtained as the detection torque value TQ2, and the judgment torque value TQ3 is calculated. The judgment torque value TQ3 is equal to the detection torque value TQ2 minus the calibration torque value TQ1, that is: TQ3 = TQ2-TQ1

[0042] The obtained judgment torque value TQ3 mainly reflects the torque generated by the driver's hands on the steering wheel.

[0043] As another embodiment, the steering wheel torque signal during the system online calibration process can also be read from the torque sensor provided on the steering wheel column of the prior art, that is, the steering wheel torque can be collected by adopting the means of collecting the steering wheel torque in the current torque-based steering wheel hands-off detection on the vehicle.

[0044] 3. Hands-off detection judgment logic.

[0045] Based on the processed torque value TQ3, the steering wheel hands-off detection module implements hands-off detection logic. By designing state machine jump logic based on the torque value TQ3, it can accurately identify whether the driver has taken their hands off the steering wheel, eliminating individual vehicle differences.

[0046] Specifically, as shown in Figure 2, when the vehicle enters the intelligent driving mode, the steering wheel hands-off detection module defaults to judging that the system enters the "hands-off" state (that is, the system initially enters the "hands-off" state); when the judgment torque value TQ3 obtained after processing is less than the first set threshold value Threhold1, it enters the "hands-off pending confirmation" state, and the state machine outputs a signal to the outside that is still a hands-off sign; when the judgment torque value TQ3 is greater than the second set threshold value Threhold2 during the waiting process in the "hands-off pending confirmation" state, the system returns to the "hands-off" state (the direction of the "hands-off pending confirmation" state 1 in Figure 2); when the waiting process duration T in the "hands-off pending confirmation" state is greater than the first time threshold t1, it is judged that the system enters the "hands-off" state (the direction of the "hands-off pending confirmation" state 2 in Figure 2), and the state machine outputs a signal that is a hands-off sign.

[0047] The logic of the system transferring from the "hands off" state to the "hands not off" state is similar. When the judgment torque value TQ3 obtained after processing is greater than the third set threshold value Threhold3, the system enters the "hands off waiting for confirmation" state. At this time, the state machine outputs a signal that is still a hands-off mark. When the judgment torque value TQ3 is less than the fourth set threshold value Threhold4 during the waiting process of the "hands off waiting for confirmation" state, the system returns to the "hands off" state (the direction of the "hands off waiting for confirmation" state 1 in Figure 2); when the waiting process duration T of the "hands off waiting for confirmation" state is greater than the second time threshold t2, the system is judged to enter the "hands off" state (the direction of the "hands off waiting for confirmation" state 2 in Figure 2), and the state machine outputs a signal that is a hands-off mark.

[0048] To avoid frequent state changes, the first threshold Threhold 1 in the "hands-on" state is lower than the second threshold Threhold 2; the third threshold Threhold 3 in the "hands-off" state is higher than the fourth threshold Threhold 4. The relationship between the first and second thresholds Threhold 1 and Threhold 2 used for determining the transition from "hands-on" to "hands-off" states, and the third and fourth thresholds Threhold 3 and Threhold 4 used for determining the transition from "hands-off" to "hands-on" states, can be set based on actual conditions and the required sensitivity, that is, a balance can be adjusted based on the missed detection rate and the false positive rate.

[0049] The first time threshold t1 and the second time threshold t2 can also be set according to actual conditions.

[0050] In step 1 of this embodiment, the acquisition of the calibrated torque value adopts a torque self-learning compensation method to solve the problem that the mechanical friction of the steering systems of different vehicles varies. When superimposed on the steering wheel torque sensor, the same software algorithm is difficult to adapt to different vehicles, resulting in uneven recognition accuracy for different vehicles after mass production.

[0051] At the same time, the hands-off detection judgment in step 3 adopts the hands-off judgment state machine strip logic to solve the problem of low accuracy and frequent state jumps of conventional hands-off detection algorithms due to the small torque change before and after the hands-off, effectively reducing false alarms and being able to balance false alarms and missed alarms by adjusting the threshold.

[0052] Method Example 2:

[0053] This embodiment differs from Method Example 1 in that, while Method Example 1 performs system online calibration before the vehicle rolls off the assembly line and leaves the factory, the method of the present invention is not limited to this. It can also be performed as a maintenance item after the vehicle is delivered and used by the user for a period of time to recalibrate steering system friction torque that has changed due to use and long-term operation. During vehicle use, especially for commercial vehicles that are used continuously for long periods under heavy loads and in harsh environments, steering system friction torque changes due to load and wear. Heavy trucks and construction vehicles, due to harsh operating environments, may also experience changes in steering damping due to dust contamination of the grease between steering mechanism components or deterioration and aging in cold, high temperature, and high humidity. These changes will ultimately be reflected in the friction torque, resulting in changes in the steering system friction torque.

[0054] The method of the present invention can also be used for recalibration when the torque-based steering wheel hands-off detection accuracy decreases due to changes in the steering system friction torque after the vehicle has been used for a period of time.

[0055] Specifically, under the condition of ensuring safety in a dedicated venue, closed road section or open road, the calibration state can be entered by the offline calibration with the same trigger method and calibration method in method embodiment 1, and the calibration torque value TQ1 can be recalibrated. After the new calibration torque value TQ1 is obtained by recalibration, the original calibration torque value is overwritten.

[0056] The acquisition and processing of the driver's hands-off detection signal during subsequent vehicle use, as well as the hands-off detection judgment logic are the same as those in Method Example 1 and will not be repeated here.

[0057] Method Example 3:

[0058] The only difference between this embodiment and method embodiment 1 is that the hands-off detection judgment logic in step 3 may also adopt other judgment logics. For example, a judgment threshold may be set more simply. When the judgment torque value TQ3 obtained after processing is less than the judgment threshold and reaches the set time, the system is judged to enter the "hands-off" state and a hands-off flag is output; if the judgment torque value TQ3 is greater than the judgment threshold within the set time, it is still in the "hands-off state" and the hands-off flag is not output.

[0059] In the "hands-off" state, if the judgment torque value TQ3 obtained after processing is greater than the judgment threshold and reaches the set time, the system is judged to enter the "hands-on" state and the hands-on flag is output; if the judgment torque value TQ3 is less than the judgment threshold again within the set time, it is still in the "hands-off" state and the hands-on flag is not output.

[0060] Alternatively, two judgment thresholds can be set: a first threshold and a second threshold that is greater than the first threshold. When the processed judgment torque value TQ3 is less than the smaller first threshold, the system is judged to have entered the "hands-off" state and a hands-off flag is output. In the "hands-off" state, when the processed judgment torque value TQ3 is greater than the larger second threshold, the system is judged to have entered the "hands-on" state and a hands-on flag is output. The second threshold is greater than the first threshold to avoid frequent and repeated jumps.

[0061] System Example:

[0062] The intelligent driving system of this embodiment is used to achieve unmanned autonomous driving or assisted driving without the need for a driver to operate the steering wheel. After the intelligent driving system takes over the vehicle, the driver's hands on the steering wheel are still required to ensure driving safety. The system is equipped with a hands-off steering wheel detection module. This steering wheel hands-off detection module detects the driver's hands-off steering wheel in the intelligent driving mode using the hands-off steering wheel detection method of the present invention. The specific hands-off steering wheel detection method has been clearly described in the method embodiment and will not be repeated in this embodiment.

[0063] Vehicle Example:

[0064] A vehicle in this embodiment has an intelligent driving function, and the intelligent driving function is implemented by an intelligent driving system introduced in the system embodiment.

Claims

1. A method for detecting the release of a steering wheel, characterized in that, During the intelligent driving process, the detection torque of the steering wheel for hands-off detection is obtained, and the hands-off detection result of the steering wheel is obtained after comparing the detection torque with the calibrated torque; the calibrated torque is obtained by collecting the steering wheel torque and performing statistical processing under the condition that the driver's hand is off the steering wheel during the intelligent driving process of the vehicle.

2. The method for detecting the release of the steering wheel according to claim 1, wherein, The comparison between the detection torque and the calibrated torque includes subtracting the calibrated torque from the detection torque to obtain a judgment torque, and comparing the judgment torque with one or more preset thresholds. At least when the judgment torque is less than a threshold, it is judged that the hands-off detection result of the steering wheel is hands-off; At least when the judgment torque is greater than a threshold, it is judged that the hands-off detection result of the steering wheel is not hands-off.

3. The method for detecting the release of the steering wheel according to claim 2, wherein When the judgment torque is less than a threshold and reaches the set time, it is judged that the hands-off detection result of the steering wheel is hands-off.

4. The method for detecting the release of the steering wheel according to claim 3, wherein, Within the set time, if the judgment torque is greater than a set threshold again, the hands-off detection result of the steering wheel remains the current not hands-off.

5. The method for detecting the release of the steering wheel according to claim 2, wherein When the judgment torque is greater than a threshold and reaches the set time, it is judged that the hands-off detection result of the steering wheel is not hands-off.

6. The method for detecting the release of the steering wheel according to claim 5, wherein Within the set time, if the judgment torque is less than a set threshold again, the hands-off detection result of the steering wheel remains the current hands-off.

7. The method for detecting the release of the steering wheel according to claim 1, wherein The calibration process for obtaining the calibrated torque is carried out before the vehicle leaves the factory; or after the vehicle has been running for a period of time.

8. The method for detecting the release of the steering wheel according to claim 1, wherein The statistical processing includes calculating the average value of the absolute values of the steering wheel torque within a certain period of time collected to obtain the calibrated torque.

9. The method for detecting the release of the steering wheel according to claim 1, wherein During the calibration process of obtaining the calibrated torque, the intelligent driving process of the vehicle reaches the set time.

10. A steering wheel detachment detection system, comprising a processor, characterized in that, The processor executes to implement the steering wheel hands-off detection method according to any one of claims 1 to 9.

11. A vehicle, characterized in that, It includes the steering wheel hands-off detection system according to claim 10.

Images