Automatic Vertical Parking Reference Trajectory Based on Improved Immune Shark Smell Optimization

Yan Chen, Gang Liu, Longda Wang, Bin XiaAlgorithms 2024 · DOI

TL;DR

This paper introduces an "Improved Immune Shark Smell Optimization" (IISSO) algorithm to generate smoother, shorter, and safer parking trajectories for automatic vertical parking (AVP). By combining cubic spline interpolation with novel immune, refraction, and Gaussian variation mechanisms, IISSO significantly outperforms existing methods in both accuracy and speed, paving the way for more robust autonomous parking systems.

What Problem Does This Paper Address?

Automatic Vertical Parking (AVP) is a crucial feature for intelligent vehicles, but optimizing the parking trajectory remains a significant challenge. Traditional methods often struggle to produce paths that are simultaneously collision-free, smooth, short, and accurate. This difficulty arises from the complex, non-linear nature of the problem, especially in confined spaces like parking garages. The authors aim to address this by developing a more sophisticated optimization model and algorithm that can generate high-quality parking trajectories, improving the overall performance and reliability of AVP systems.

Methods

The paper proposes a two-pronged approach:

  1. Cubic Spline Interpolation for Trajectory Modeling: The authors utilize cubic spline interpolation to construct a flexible and smooth reference trajectory model. This method allows for the generation of continuous and differentiable paths, which are essential for avoiding jerky movements during parking. The key points of the vehicle's coverage area, as detected by sensors, are used as control points for the spline.
  2. Improved Immune Shark Smell Optimization (IISSO): This is the core algorithmic contribution. IISSO is an enhancement of the original Shark Smell Optimization (SSO) algorithm. It incorporates three key mechanisms to improve global optimization capabilities and escape local optima:
  3. Immune Mechanism: Inspired by the human immune system, this mechanism uses an "antigen" (global optimum) and "antibodies" (individuals in the population) to adaptively select and promote better solutions, maintaining population diversity.
  4. Refraction Mechanism: This mechanism simulates light refraction. When a solution gets trapped in a local optimum, it generates "refracted" solutions that are more likely to lead to the global optimum.
  5. Gaussian Variation Mechanism: This adds a controlled random perturbation to the solutions, allowing for finer local searches and helping to escape shallow local optima.

The IISSO algorithm is then used to optimize the parameters of the cubic spline trajectory, aiming to minimize path length while satisfying various constraints like collision avoidance and parking inclination angle.

Datasets

The paper describes experimental verification conducted in real-world parking scenarios. Two specific AVP experiment scenes are detailed:

  • Scene 1: A garage in Dalian Shell Museum, using a Toyota LeiLing Shuangqing 185T Sportline (2020 model).
  • Scene 2: Another garage in Dalian Shell Museum, using a Toyota Corolla 1.2T S-CVT GL Pioneer Version (2019 model).

These are not datasets in the traditional sense of pre-collected data for training. Instead, they represent the physical environments and vehicles used for experimental validation of the proposed algorithm.

Key Insights and Ideas

  • Synergy of Splines and Swarm Intelligence: The paper highlights the effectiveness of combining a smooth trajectory generation method (cubic splines) with a powerful optimization algorithm (IISSO) to tackle the complex AVP problem.
  • Multi-Mechanism Optimization: The IISSO's strength lies in its integration of multiple strategies (immune, refraction, Gaussian variation). This ensemble approach provides a more robust way to explore the search space and avoid getting stuck in local optima, which is a common pitfall for many optimization algorithms.
  • Balancing Exploration and Exploitation: The immune and refraction mechanisms seem to focus on global exploration, while Gaussian variation aids in local exploitation, creating a well-balanced optimization process.
  • Practical Relevance: The experiments conducted with real vehicles in actual parking garages demonstrate the practical applicability and effectiveness of the proposed IISSO algorithm.

Key Results

The experimental results, presented in Tables 2-5 and Figures 5-12, show that the proposed IISSO algorithm consistently outperforms other compared algorithms (IIMFO, SSO, MFO, IPSO, PSO) in several key metrics:

  • Shorter Path Length: IISSO achieved the shortest reference trajectory path length in both experimental scenes.
  • Lower Parking Position Error: The final parking position accuracy was improved, with IISSO resulting in smaller errors.
  • Reduced Parking Inclination Angle: The algorithm generated trajectories with a lower actual parking inclination angle, contributing to a smoother and more stable parking maneuver.
  • Faster Optimization: IISSO demonstrated a faster realization time for optimization compared to most other algorithms.
  • Improved Tracking Control: The tracking control curves also showed that IISSO's optimized trajectories led to more accurate and efficient parking maneuvers.

Limitations and Future Work

The authors acknowledge several limitations:

  • Dynamic Obstacles: The current research assumes a static parking environment. Real-world parking often involves dynamic obstacles (pedestrians, other vehicles). While the system can stop for sudden obstacles, it lacks the ability to continue tracking and avoiding them.
  • Vehicle Type: The research was limited to ordinary vehicles. It is not directly applicable to special vehicles like trucks or large vehicles, which have different kinematic constraints.
  • Further Improvement: While the results are improved, there is still room for further enhancement of the algorithm's performance.
  • System Complexity and Cost: The experimental AVP system is relatively complex and expensive, indicating that mass production of such systems is still some way off.

Future work could focus on incorporating dynamic obstacle avoidance, extending the algorithm to handle a wider range of vehicle types, and further refining the optimization mechanisms.

Summary updated

Abstract

Parking path optimization is the principal problem of automatic vertical parking (AVP); however, it is difficult to determine a collision avoiding, smooth, and accurate optimized parking path using traditional parking reference trajectory optimization methods. In order to implement high-performance automatic parking reference trajectory optimization, we establish an automatic parking reference trajectory optimization model using cubic spline interpolation, and we propose an improved immune shark smell optimization (IISSO) to solve it. Firstly, we take the length of the parking reference trajectory as the optimization objective, and we introduce an intelligent automatic parking path optimization model using cubic spline interpolation. Secondly, the improved immune shark optimization algorithm combines the immune, refraction, and Gaussian variation mechanisms, thus effectively improving its global optimization ability. The simulation results for the parking path optimization experiments indicate that the proposed IISSO has a higher optimization accuracy and faster calculation speed; hence, it can obtain a parking path with higher optimization performance.
BibTeX
@inproceedings{chen2024automatic,
  author = {Yan Chen and Gang Liu and Longda Wang and Bin Xia},
  title = {{Automatic Vertical Parking Reference Trajectory Based on Improved Immune Shark Smell Optimization}},
  booktitle = {Algorithms},
  year = {2024},
  doi = {10.3390/a17070308},
}
Figure