Next-Park-AI: A Smart Parking Prediction and Recommendation System

The reviewed literature shows that machine-learning and deep-learning techniques — ANN/ARNN-based destination prediction, Random Forest-based occupancy classification, and LSTM-based violation forecasting — have each demonstrated strong, measurable improvements in specific aspects of parking management, yet no single existing system combines real-time availability, AI-driven recommendation, online booking, secure payment, and compliance verification into one integrated platform. Next-Park-AI is being developed to close this gap by combining a MERN-stack web application with an AI/ML recommendation and prediction module, real-time synchronization, QR-based verification, and integrated payment. The system is expected to improve space utilization, reduce congestion and search time, and deliver a more efficient parking experience for both users and administrators, with future scalability toward smart cities, malls, airports, hospitals, and other commercial parking facilities. That said, this review has its own limitations, and it is worth being upfront about them rather than overstating what has actually been established. First, the comparative analysis in Sections 3 and 4 is built on three core papers plus a handful of supporting citations; it is not an exhaustive survey of the smart-parking literature, and other relevant approaches — particularly newer deep-learning or transformer-based methods — may exist outside its scope. Second, the design decisions drawn from that analysis, including the choice of Random Forest for occupancy prediction and a weighted-scoring approach for slot recommendation, are reasoned from the reported results of other systems rather than from any experiments run on Next-Park-AI's own data, since the platform has no historical booking history to train or validate against yet. Third, the accuracy figures and compliance findings quoted throughout this paper (97.60% destination accuracy, 93–94% availability accuracy, a KS statistic of 0.916) belong to the original studies and their specific datasets — Dublin micromobility trips, Birmingham IoT sensors, and a single US university campus — and cannot be assumed to transfer directly to Next-Park-AI's deployment context without independent evaluation once the system is live. Finally, the current design does not incorporate physical IoT occupancy sensors, weather data, or trip-origin information, relying instead on booking-based occupancy records; extending the model with these inputs is left as future work once the platform has real usage data to draw on.

[1]   S. Yan, N. E. O'Connor, and M. Liu, "U-Park: A User-Centric Smart Parking Recommendation System for Electric Shared Micromobility Services," IEEE Transactions on Artificial Intelligence, 2024. DOI: 10.1109/TAI.2024.3428513. [2]   A. Dahiya, P. Mittal, Y. K. Sharma, U. K. Lilhore, S. Simaiya, E. Ghith, and M. Tlija, "Machine Learning-Based Prediction of Parking Space Availability in IoT-Enabled Smart Parking Management Systems," Journal of Advanced Transportation, vol. 2024, Article ID 8474973, 2024. DOI: 10.1155/2024/8474973. [3]   A. Javaheri, S. S. Channamallu, S. Kermanshachi, J. M. Rosenberger, A. Pamidimukkala, C. Kan, and G. Hladik, "Evaluating Impact of Smart Parking Systems on Parking Violations," IEEE Access, 2024. DOI: 10.1109/ACCESS.2024.3503513. [4]   A. O. Kotb, Y. Shen, X. Zhu, and Y. Huang, "The Role of Smart Parking in Building Smart Cities," Journal of Management Analytics, vol. 5, no. 4, pp. 265-284, 2018. [5]   S. Jiang, W. Deng, and L. Wang, "Design and Implementation of a Smart Parking System," IEEE Transactions on Intelligent Transportation Systems, vol. 21, no. 8, pp. 3423-3430, 2020. [6]   S. Saharan, N. Kumar, and S. Bawa, "DyPARK: A Dynamic Pricing and Allocation Scheme for Smart On-Street Parking System," IEEE Transactions on Intelligent Transportation Systems, vol. 24, no. 4, pp. 4217-4234, 2023.