Abstract:Autonomous driving is among the most promising of upcoming traffic safety technologies. Prototypes of autonomous vehicles are already being tested on public streets today. However, while current prototypes prove the feasibility of truly driverless cars, edge cases remain which necessitate falling back on human operators. Teleoperated driving is one solution that would allow a human to remotely control a vehicle via mobile radio networks. Removing in-vehicle drivers would thus allow current autonomous technologies to further progress towards becoming genuinely driverless systems. This paper proposes a new model predictive steering control scheme, specifically designed for semi-autonomous, teleoperated road vehicles. The controller is capable of receiving teleoperator steering commands and, in the case of potential collisions, automatically correcting these commands. Collision avoidance is incorporated into the design using potential fields. A term in the cost function facilitates natural maneuvers, and constraints on the maximum potential keep the vehicle at safe distances from obstacles. This paper also proposes the use of high-order ellipses as a method to accurately model rectangular obstacles in tight driving scenarios. Simulation results support the effectiveness of the proposed approach.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is how to ensure the safe driving of remotely - controlled semi - autonomous vehicles in complex traffic environments and be able to automatically correct the steering instructions of human operators when necessary to avoid collisions. Specifically, the article proposes a new model based on predictive control and the potential field method, aiming at:
1. **Tracking the steering instructions of remote operators**: Ensure that the vehicle can drive according to the intentions of the remote operator.
2. **Automatically correct steering instructions to avoid collisions**: When a potential collision risk is detected, the controller can automatically adjust the steering instructions to make the vehicle avoid obstacles.
3. **Improve the modeling accuracy of rectangular obstacles**: Use high - order ellipses to represent rectangular obstacles (such as other vehicles) more accurately, so as to better adapt to narrow driving scenarios.
### Specific problem description
1. **Challenges in remote - control driving**:
- One of the main challenges in remote - control driving is the latency problem, that is, there is a time difference between the time when instructions are sent from the remote control center and the time when the vehicle executes the instructions. If the latency is too large, the vehicle may not be able to respond in time.
- Another challenge is the insufficient situation awareness of remote operators. Since the operator is not in the vehicle, additional attention is required to compensate for the distortion of sensor data or missing information.
2. **Limitations of existing methods**:
- Traditional remote - control methods may not be able to fully handle complex traffic situations, especially in emergency situations, which may lead to collisions between vehicles and obstacles.
- Existing obstacle - avoidance methods mostly rely on spatial constraints and are difficult to flexibly deal with obstacles of different shapes and sizes.
### Proposed solutions
1. **Predictive control based on the potential field method**:
- The potential field method guides the vehicle to avoid obstacles by introducing attraction fields and repulsion fields. The repulsion field is used to represent obstacles, and the attraction field is used to guide the vehicle towards the target path.
- In the potential field function \( P(x, y)=\frac{\alpha}{(s_q(x, y)+ 1)^\beta} \), the parameters \(\alpha\) and \(\beta\) control the intensity and slope of the potential field respectively, and \( s_q(x, y) \) describes the shape of the obstacle.
2. **Modeling rectangular obstacles with high - order ellipses**:
- Use the high - order ellipse \( s(x, y)=\left(\frac{x - x_e}{a}\right)^n+\left(\frac{y - y_e}{b}\right)^n - 1 \) to represent rectangular obstacles more accurately, where \( n \) is the order of the ellipse, and \( a \) and \( b \) are the semi - axis lengths of the ellipse.
- By adjusting the order \( n \) of the ellipse, rectangular obstacles can be modeled more accurately in narrow driving scenarios, reducing unnecessary buffer zones.
### Model predictive control (MPC) formulation
To achieve the above goals, the author formulates the control problem as a model predictive control (MPC) problem. The objective function of the optimization problem consists of three parts:
1. **Tracking the steering instructions of remote operators**:
\[
J_{\delta_{\text{ref}}}(\delta_0)=w_{\delta_{\text{ref}}}(\delta_0-\delta_{\text{ref}}(t))^2
\]
where \(\delta_0\) is the input of the controller, \(\delta_{\text{ref}}(t)\) is the reference input of the remote operator, and \(w_{\delta_{\text{ref}}}\) is the weight coefficient.
2. **Obstacle - avoidance potential field**:
\[
J_P(\eta_N)=w_P\sum_{i = 1}^N\left(P(x_{\text{fl}, i}, y_{\text{fl}, i})+P(x_{\text{fr}, i}, y_{\text{fr}, i})\right)
\]
where \(\eta_N\) is the system output.