Method for sensing and determining control information for a vehicle using a wireless communication channel

Abstract

The invention relates to method for sensing and determining control information (216) for a vehicle (110a, 110b) using a wireless communication channel (120), wherein a signal frame (130a, 130b) sent via the wireless communication channel comprises a sensing signal (132) and a control signal (134), the method comprising, in a sensing and control cycle to be repeated: Providing (206) position information (204), comprising information about a position of the vehicle (110a, 110b), the position having been determined based on a response signal that has been received in response to the sensing signal of a signal frame having been sent to the vehicle; Determining (214) control information (216) for the vehicle, based on an accuracy (212) of the position of the vehicle, and on context information (218); Determining (220) fractions (132a, 134a) of the sensing signal and the control signal for a next signal frame, by means of a machine learning algorithm (222); Causing (224) the next signal frame to be sent via the wireless communication channel (120) to the vehicle (110a, 110b), wherein the control signal of the next signal frame is based on or comprises the control information.

Description

[0001] R.414510

[0002] Description

[0003] Title

[0004] Method for sensing and determining control information for a vehicle using a wireless communication channel

[0005] The present invention relates to a method for sensing and determining control information for a vehicle using a wireless communication channel, wherein a signal frame sent via the wireless communication channel comprises a sensing signal of the vehicle and a control signal, to a computing device and a computer program for performing the method.

[0006] Invention background

[0007] Vehicles like automated guide vehicles (AGVs), robots or drones (un-manned aerial vehicle) can be controlled using a wireless communication channel, which is e.g., based on 5G or 6G. Such wireless communication channel can not only be used for controlling the vehicle, i.e. , transmitting control information according to which the vehicle moves but also for sensing the vehicle, i.e., determining a position of the vehicle.

[0008] Disclosure of the invention

[0009] According to the invention, a method for sensing and determining control information for, specifically for controlling, a vehicle, a computing device and a computer program having the features of the independent claims are proposed. Advantageous embodiments are subject of the dependent claims and the following description. R.414510

[0010] - 2 -

[0011] The invention generally relates to controlling movement of vehicles via a wireless communication channel (i.e. , based on a wireless communication network). A preferred kind of wireless communication network to be used is a mobile wireless communication network (cell phone networks) according to 3GPP standards such as 5G or 6G. There are different kinds of vehicles that can be controlled in this way, in particular, unmanned vehicles like robots, AGVs (automated guided vehicles) or drones, or other remotely controlled vehicles.

[0012] Such wireless communication channel can not only be used for controlling the vehicle, i.e., by sending control information to the vehicle, but also for sensing the vehicle. Sensing the vehicle means determining (or sensing) the position of the vehicle using the wireless communication channel. In order to combine both, signal frames to be sent via the wireless communication channel can be used, where a signal frame comprises a sensing signal and a control signal. The sensing signal is, in particular, a signal to be sent to the vehicle and to be reflected on a surface of the vehicle, preferably using radio resources of the wireless communication channel; based on the reflected signal, the position can be determined, e.g., like with RADAR signals. The control signal comprises or is based on control information which is transmitted to the vehicle in this way. The control signal is received by the vehicle (or a receiving module thereof), such that the control information comprised therein can be processed in order to control the vehicle. The control information comprises, for example, information about a velocity and / or a heading direction for the vehicle.

[0013] In joint communication sensing and control, a base station uses the wireless medium, i.e., the wireless communication channel, for sensing and for communication (including control). In sensing stage, the base station sends sensing signal to detect the vehicle positioning. Based on the sensing feedback, i.e., the response signal that has been received in response to the sensing signal, the base station has information about the vehicle positioning and then it can internally apply its path tracking algorithm, for example; typically, a trajectory is given for the vehicle, which is to follow. Based on the control signal calculated, the base station sends it (in a next signal frame) to the vehicle through the wireless communication channel, aiming to keep the vehicle on track and avoid R.414510

[0014] - 3 - path deviation. Different vehicles and respective control systems (e.g., AGVs or drones) typically need a different level of communication (control) and sensing. Moreover, such levels also depend on the task of the vehicle itself. This means that the fractions of the sensing signal and the control signal for a signal frame are different.

[0015] For a given limited amount of network resources, there is a trade-off between the sensing capacity and the communication (control) capacity. So, if a better accuracy at the vehicle positioning is required or wanted, more network resources have to be used for sensing (bigger fraction for the sensing signal), thus potentially affecting the communication (control) capacity (e.g., throughput). On the other hand, for more communication (control) capacity, the positioning accuracy is affected.

[0016] Within the present invention, a way to improve how to choose the fractions of sensing signal and control signal is presented. For operating the vehicle, a sensing and control cycle is, e.g., continuously, repeated. In such sensing and control cycle, position information is provided, which comprises information about a position of the vehicle, the position having been determined based on a response signal that has been received in response to the sensing signal of a signal frame having been sent to the vehicle; this signal frame can have been sent within the previous sensing and control cycle.

[0017] Based on an accuracy of the position of the vehicle, and based on context information, control information for the vehicle is determined. In addition, the control information can also be determined based on wireless control information. The context information comprises, in an embodiment, information about at least one of the following parameters: a reference state, an intended task for the vehicle, an environment map, an operational parameter for the vehicle. The reference state can be the desired physical state. For example, the trajectory the robot should follow (reference state - planned trajectory) compared to its current trajectory (current state). The operational parameters can comprise, for example, a steering angle for the wheels, aceleration, force, etc. The wireless communication information can comprise information about wireless R.414510

[0018] - 4 - communication capacity, bandwidth, power, interference or other key performance indicators (KPIs). The control information can comprise information about a velocity and / or a heading direction for the vehicle. The accuracy of the position of the vehicle can be determined based on the position of the vehicle and a given trajectory for the vehicle. In this way, the control information can be such that the vehicle is to be controlled to further follow the given trajectory or get back onto it, for example, also considering current needs.

[0019] Further, fractions of the sensing signal and the control signal for a next signal frame are determined, by means of a machine learning algorithm. In particular, the fractions of the sensing signal and the control signal for the next signal frame are determined based on information having been determined in a or multiple previous sensing and control cycles, i.e. , on historic data. This information having been determined in each of the previous sensing and control cycles comprises: the accuracy of the position of the vehicle, the context information, and the control information. This can also comprise the wireless communication information. Further, this can include the history of wireless communication information, such as transmission errors, block error rate (BLER), etc.

[0020] The machine learning algorithm may be configured to determine a first fraction of the next signal frame for the sensing signal and a second fraction of the next signal frame for the control signal. The first and the second fraction correspond to radio resources which may differ in the time and / or the frequency domain, e.g., different (sub-)slots and / or different (sub-)carriers. Here, the machine learning algorithm is preferably configured to receive as input data at least one of the accuracy of the position of the vehicle, the context information, the control information, as having been determined in the one or multiple previous sensing and control cycles. The machine learning algorithm is preferably further configured to determine output data based on the received input data, wherein the output data represent the fractions of the sensing signal and the control signal for the next signal frame, e.g., a split of the next signal frame into a first and a second fraction corresponding to the sensing signal and the control signal, respectively. R.414510

[0021] - 5 -

[0022] The machine learning algorithm may be trained with training data representing a relation between a split of the signal frame into fractions for the sensing control and the control signal and corresponding accuracies of the position of the vehicle and / or context information and / or control information.

[0023] In addition, within or after each sensing and control cycle, the machine learning algorithm is adapted, based on the accuracy of the position of the vehicle, the context information, the control information having been determined in the sensing and control cycle. It can, in addition, be adapted based on the wireless communication information.

[0024] Further, the next signal frame is caused to be sent via the wireless communication channel to the vehicle, e.g., by initiating a transmission and / or sending control information to a sending module. The control signal of the next signal frame is based on or comprises the control information.

[0025] To sum up, a principle of the present invention is a method for balancing the sensing and communication trade-off, where a learning algorithm is applied to evaluate the critical states of the control system and, based on that, balances the communication (control) and sensing. This can be considered a holistic approach that seamlessly integrates communication, sensing, and control functions within a single frame or framework, dynamically optimizing the trade-off between these critical functions in real-time. By employing an adaptive learning algorithm, the proposed way intelligently allocates network resources based on the immediate requirements of the vehicle’s operational context. This ensures optimal performance across all three domains - enhancing the vehicle’s situational awareness through precise sensing, ensuring robust and responsive control through efficient communication, and dynamically adjusting these parameters as the operational environment or objectives change. This approach not only solves the pressing problem of resource allocation inefficiency but also propels the capabilities of mobile vehicle or robotic systems to new heights, fully leveraging R.414510

[0026] - 6 - the potential of in particular 6G technology to meet the demands of future smart environments.

[0027] A computing device according to the invention (a system for data processing), e.g., a control device or a control unit of a base station, or a central server or other computing system, is configured, in particular programmatically, to perform a method according to the invention.

[0028] In an embodiment, the vehicle comprises the computing device. In another embodiment, the vehicle is configured to receive an RIS configuration for a reconfigurable intelligent surface, RIS. In the latter case, the determination of the RIS configuration can take place in the cloud (or a central server or the like), to which the jamming signal information is sent, e.g., by the vehicle or a communication module thereof. The final configuration of the RIS is then also caused by the vehicle or a corresponding module.

[0029] The implementation of a method according to the invention in the form of a computer program or computer program product with program code for performing all method steps is also advantageous, since this causes particularly low costs, especially if an executing control unit is still used for further tasks and is therefore present anyway. Finally, a machine-readable storage medium is provided with a computer program stored thereon as described above. Suitable storage media or data carriers for providing the computer program are, in particular, magnetic, optical and electrical memories, such as hard disks, flash memories, EEPROMs, DVDs and the like. It is also possible to download a program via computer networks (Internet, intranet, etc.). Such a download can take place wired or wirelessly (e.g., via a WLAN network, a 3G, 4G, 5G or 6G connection, etc.).

[0030] Further advantages and embodiments of the invention will be apparent from the description and the accompanying drawing.

[0031] The invention is shown schematically in the drawing by means of an example of an embodiment and is described below with reference to the drawing. R.414510

[0032] - 7 -

[0033] Brief description of the drawings

[0034] Fig. 1a schematically shows an arrangement with vehicles to explain the invention.

[0035] Fig. 1b schematically shows signal frames to explain the invention.

[0036] Fig. 2 schematically shows a method according to an embodiment.

[0037] Embodiment(s) of the invention

[0038] Fig. 1a schematically illustrates an arrangement with vehicles to explain the invention. There is a base station 100 comprising a computing device 102 and a communication module 104. Further, there is a vehicle 110a, which is an AGV in this embodiment. The AGV 110a comprises a computing device 112 and a communication module 1114. Further, there is a vehicle 110b, which is a drone in this embodiment. The drone 100b can also comprise a computing device and a communication module.

[0039] For controlling AGV 110a (or its movement), base station 100 can repeatedly send signal frames via wireless communication channel 120 to AGV 110a. Similarly, for controlling drone 110b, base station 100 can repeatedly send signal frames via a wireless communication channel. Note that a base station does not necessarily have to be able to control both, an AGV and a drone; this is only for illustration purposes and to explain the invention.

[0040] In Fig. 1b, signal frames are illustrated, showing a frequency f vs. time t. Signal frame 130a has a certain length in time, and comprises a sensing signal 132 and a control signal 134. Sensing signal 132 has a fraction 132a of the entire signal frame, whereas control signal 134 has a fraction 134a of the entire signal frame. Signal frame 130b is similar to signal frame 130a, however, its sensing signal and control signal have different fractions. R.414510

[0041] - 8 -

[0042] For example, signal frame 130a can be used to sense and control AGV 110a, where sensing the AGV’s position is less important than controlling its exact position. Thus, the fraction of the sensing signal is low compared to the fraction of the control signal. Signal frame 130b, for example, can be used to sense and control done 110b, where sensing the drone’s position is more important than controlling its exact position. Thus, the fraction of the sensing signal is high compared to the fraction of the control signal.

[0043] Fig. 2 schematically illustrates a method for sensing and determining control information for a vehicle using a wireless communication channel according to an embodiment. Such method can be implemented in the situation as shown in Figs. 1a, 1 b, for example, for the AGV or the drone.

[0044] In step 200, a response signal to the sensing signal of a signal frame having been sent to the vehicle, e.g., in the previous cycle, is received. Based on this response signal, step 202, a position of the vehicle is determined. In step 206, position information 208 is provided, the position information comprising information about the position of the vehicle.

[0045] In step 210, an accuracy 212 of the position is determined, based on the position itself and a given trajectory for the vehicle. In step 214, control information 216 for the vehicle is determined, based on the accuracy 212 of the position of the vehicle, and on context information 218. The context information 218 can comprise information about at least one of the following parameters: a reference state, an intended task for the vehicle, an environment map, an operational parameter for the vehicle. The environment map can, for example, also include information about obstacles in the environment, which are to be observed by the vehicle.

[0046] In step 220, fractions of the sensing signal and the control signal for a next signal frame are determined, by means of a machine learning algorithm 222. This is, in particular, based on information having been determined in previous sensing and control cycles, the information having been determined in each of the previous sensing and control cycles comprising: the accuracy of the position of the vehicle, R.414510

[0047] - 9 - the context information, and the control information. Two examples for such fractions are shown in Fig. 1b.

[0048] In step 224, the next signal frame is caused to be sent (and then is sent) via the wireless communication channel to the vehicle. The control signal of the next signal frame is based on or comprises the control information 216.

[0049] These steps can be repeated in a cycle; this means, for example, that the next signal frame mentioned, results in a new response signal that is received, like in step 200.

[0050] It is noted that in a very first signal frame, the fractions can be chosen e.g., according to experience, other requirements or simply to be equal. By a learning stage the algorithm is continuously refined through experience. Each cycle of decision-making, feedback, and command execution enriches the understanding and capability. This iterative process serves as the foundation for developing an optimal policy for how to determine the fractions. Such a policy is designed to dynamically allocate network resources for communication and sensing in a manner that is in perfect harmony with the control commands and contextual information of the control system.

[0051] If, for example, if for a specific situation according to the context data, in previous cycles the fraction of the sensing signal was low and had to be increased due to low accuracy of the position, for a next signal frame the fraction of the sensing signal can immediately be chosen higher.

[0052] The proposed method directly confronts a pervasive issue in the domain of mobile robotics or vehicles, especially in environments where Automated Guided Vehicles (AGVs) operate - namely, the optimal allocation of limited wireless network resources to balance precise sensing and effective communication and controlling. This dilemma becomes particularly evident in scenarios where the requirements for sensing and communication fluctuate significantly due to changes in the operational context or the tasks being performed, e.g., by the AGVs. R.414510

[0053] - 10 -

[0054] For instance, in a factory floor scenario where an AGV is tasked with transporting materials from one station to another, the AGV navigates through crowded and dynamically changing environments, and precise positioning becomes paramount to avoid collisions with obstacles or other AGVs. In this context, allocating a larger share of network resources to sensing ensures the AGV's accurate realtime localization, but at the cost of reducing the bandwidth available for communication (controlling). This reduction can hinder the AGV's ability to receive timely control signals, potentially leading to inefficiencies or operational delays.

[0055] Conversely, in a scenario where the AGV needs to execute high-speed maneuvers or coordinate closely with other AGVs to perform a task, the priority shifts towards ensuring robust and high-throughput communication. Here, communication (controlling) bandwidth allocation might be favoured, potentially compromising the AGV's positioning accuracy. Such a compromise could result in less precise movements or difficulty in maintaining optimal paths, impacting overall task efficiency and safety.

[0056] This specific problem of dynamically balancing the trade-off between sensing and communication is not adequately addressed by existing systems, which often statically allocate network resources without considering the changing operational demands or environmental contexts. The inability to adaptively allocate resources based on real-time needs leads to suboptimal performance, reduced efficiency, and increased operational risks for mobile robotic systems.

[0057] The proposed adaptive learning methodology empowers a base station (or other control system) to dynamically balance the critical needs of positioning accuracy (sensing) and communication capacity (controlling) for mobile vehicle or robotic systems, like AGVs, in particular in a 6G wireless communication context.

[0058] This balance can be crucial for optimizing the control performance of these systems across various operational scenarios. By employing machine learning algorithms, the base station analyses real-time data on the vehicle’s position and R.414510

[0059] - 11 - movement, learning to identify critical states of the control system where either sensing or communication demands peak. Depending on the identified need - be it enhanced positioning for precise maneuvering or increased communication for complex task coordination - the system intelligently adjusts resource allocation to ensure optimal performance. This dynamic, context-aware approach significantly improves the efficiency, safety, and reliability of mobile vehicle or robotic operations, marking a significant advancement in the integration of communication, sensing, and control technologies for the future of smart environments.

[0060] The proposed method uses a learning algorithm, which p can take various inputs, such as the history of decision-making, the history of states in the vehicle application (including position accuracy and speed), and contextual data (such as the presence of obstacles). It also considers the criticality of control commands and reference states. Ideally, the learning algorithm should have access to comprehensive details of the vehicle’s control system, including control system states, current control actions and their criticality, scenario details, and the decision-making history. However, even with limited information, the algorithm can still achieve (probably sub-optimal but better than up to now) solutions by obtaining only the current decision-making action and the vehicle position accuracy.

[0061] This proposed way heralds several significant improvements over existing solutions in the realm of wireless communication and mobile robotics. Foremost, it introduces an adaptive learning-based approach to dynamically balance sensing and communication resources, ensuring optimal allocation tailored to real-time operational demands. Unlike static allocation strategies that do not account for changing environmental conditions or task requirements, this methodology significantly enhances the precision in vehicle or robot positioning and the efficiency of control signal transmission. Moreover, by incorporating context information (or context data) into the decision-making process, the proposed way offers a nuanced understanding of operational scenarios, leading to more informed and effective resource management decisions. The development of an optimal decision making policy (determining the fractions of R.414510

[0062] - 12 - the sensing and control signal), informed by continuous feedback and learning from the system's performance, represents a leap forward in achieving higher operational efficiency, safety, and reliability for vehicles like mobile robots. This approach not only optimizes network resource utilization but also sets a new standard for adaptability and performance in the integration of communication, sensing, and control technologies, particularly in the emerging landscape of 6G wireless communication.

Claims

R.414510- 13 -Claims1. A method for sensing and determining control information (216) for a vehicle (110a, 110b) using a wireless communication channel (120), wherein a signal frame (130a, 130b) sent via the wireless communication channel (120) comprises a sensing signal (132) and a control signal (134), the method comprising, in a sensing and control cycle to be repeated:Providing (206) position information (204), comprising information about a position of the vehicle (110a, 110b), the position having been determined based on a response signal that has been received in response to the sensing signal (132) of a signal frame having been sent to the vehicle (110a, 110b);Determining (214) control information (216) for the vehicle (110a, 110b), based on an accuracy (212) of the position of the vehicle (110a, 110b), on context information (218), and, preferably, on wireless communication information;Determining (220) fractions (132a, 134a) of the sensing signal (132) and the control signal (134) for a next signal frame, by means of a machine learning algorithm (222);Causing (224) the next signal frame to be sent via the wireless communication channel (120) to the vehicle (110a, 110b), wherein the control signal of the next signal frame is based on or comprises the control information (216).

2. The method of claim 1 , wherein determining (220) the fractions (132a, 134a) of the sensing signal (132) and the control signal (134) for the next signal frame is based on information having been determined in a or multiple previous sensing and control cycles, the information having been determined in each of the previous sensing and control cycles comprising: the accuracy of the position of the vehicleR.414510- 14 -(110a, 110b), the context information, the control information (216), and preferably the wireless communication information.

3. The method of claim 1 or 2, further comprising, within or after each sensing and control cycle:Adapting the machine learning algorithm (222), based on the accuracy of the position of the vehicle (110a, 110b), the context information (218), the control information (216), and preferably the wireless communication information, having been determined in the sensing and control cycle.

4. The method of any one of the preceding claims, wherein the control information (216) comprises information about a velocity and / or a heading direction for the vehicle (110a, 110b).

5. The method of any one of the preceding claims, wherein the context information (218) comprises information about at least one of the following parameters: a reference state, an intended task for the vehicle (110a, 110b), an environment map, an operational parameter for the vehicle (110a, 110b).

6. The method of any one of the preceding claims, wherein determining (214) the control information (216) for the vehicle (110a, 110b) is based on the wireless communication information, and wherein the wireless communication information comprises information about at least one of the following parameters: wireless communication capacity, bandwidth, power, interference.

7. The method of any one of the preceding claims, wherein the sensing signal (132) is a signal to be sent to the vehicle (110a, 110b) and to be reflected on a surface of the vehicle (110a, 110b).

8. The method of any one of the preceding claims, wherein the wireless communication channel is based on a mobile communication network, preferably, 5G or 6G.R.414510- 15 -9. A computing device comprising a processor configured to perform the method of any one of the preceding claims.

10. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to perform the method of any one of claims1 to 8.

11. A computer-readable medium on which the computer program of claim 10 is stored.

Images