Annual: 2019

EM022 »
ParkMe – car park guidance reinvented
📁Internet of Things
👤Kuba Kowalczykowski
 (Poznan University of Technology)
📅Oct 06, 2019
Regional Final



👀 2498   💬 3

EM022 » ParkMe – car park guidance reinvented

Description

In the era of constantly growing traffic we are challenged to find the best solutions not only for car movement organization but also for adequate parked car management. Therefore, we have to develop bigger and more complicated parking spaces than ever before. Facing these infrastructures can be tricky for many drivers and they can feel overwhelmed by car parks complexity. Innovative IoT solutions come to help these drivers who are straying in the darkness of underground car parks.

ParkMe is a simple idea for managing car park traffic based on video processing, routing, guiding and tracking of free parking places. ParkMe’s main functionality is to guide every single car individually from the car park’s entrance to a park place most suitable to driver’s needs and then all the way back to the least occupied exit.

We value the time above everything else. ParkMe is a time-saving solution for all complex parking infrastructures. With our project we would like to put an end to: - crawling around a parking & searching for a parking place, - traffic jams in the parking areas, - that irritating feeling when someone takes a place that you just spotted. Our solution will save a lot of driver’s time, which can be spent on doing something more constructive than being stuck in a traffic jam.

ParkMe’s infrastructure is based on sophisticated nodes called GuideNests, which contain Terasic FPGA boards with Intel OpenVINO Toolkit for image processing, sensors, indicators and connectivity with central server and other existing systems. GuideNests let the system individually identify every car in the car park. With the support of traffic monitoring and free parking places indicators the system can calculate the route for every driver, considering changeable environment of the car park. The system is scalable and can be expanded due to specific implementation case.

With enough time a human being is capable to do anything. Don’t waste it for parking traffic!

Demo Video

  • URL: https://youtu.be/ZkwoKCHHuVU

  • Project Proposal

    1. High-level Project Description

    Purpose of the design

    ParkMe is an concept of a system for managing car park traffic. It’s main functionality is to guide every car individually from the car park’s entrance to a parking place chosen by a driver and then all the way back to the exit - as fast as possible. 

    Although there are solutions for car parks guiding and monitoring on the market, they are rather primitive and inaccurate. Our solution will differ - thanks to video recognition we would like to obtain highest possible performance, accuracy and the most important - individuality. Each car will have its parking spot assigned at the entrance and it will be guided directly to its destination.

    Application scope

    For our project we assumed basic and supplementary goals. Our main goal is to guide an entering car individually to its destination parking place, which is chosen by driver at the entrance, regarding the current traffic. The further goals are to provide a fully scalable system for big car parks (this includes: alarms and alerts, parking places for disabled people, ambulances, armored bank cars, etc.) and wide-range server side statistics.

    Detailed description of the system usage

    1. A driver entering a car park tells the system (p.e. by a touch screen at the barrier) where he/she wants to go. He/she can choose from certain options like “food court”, “shop A”, “shop B”, ... , ”anywhere close to the entrance”. Additional button would inform the system that the driver is a disabled person to provide a route to a disabled parking place. The car’s license plate and key features are identified by the vision system.

    2. The system finds a free parking place as close to the desired building entrance as possible, considering a free parking places map and traffic map, provided by the server. 

    3. The information about the destination, connected with the car’s license plate, is stored on the server. The parking place is reserved for the particular car and cannot be reserved for another one.

    4. Route computation is done in the GuideNest and an adequate direction for the particular driver is shown on the indicator.

    5. The driver drives through the parking. Further GuideNests are recognizing the car by the license plate and key features and the adequate signs are shown on the indicators. The route calculations are based on up-to-date routing maps from the server. In case of straying from the route, a car’s destination is known by the system and it’s route can be recalculated. Every occurance of a car in certain point is posted from the GuideNest to the server with a timestamp, which lets the server to compute the traffic load.

    6. The driver reaches his park place, parks the car and goes to the desired destination. The system recognizes the car as “parked” if it enters a certain alley of the car park and does not reach next GuideNest in certain time (a bit longer than the average time of passing this alley).

    7. After returning to the car and entering the traffic the car reaches the first GuideNest from its parking place. It is recognized as “leaving” and the system guides it to the nearest exit or one of further exits if an excessive traffic occured in the nearest one. This solution is based on a presumption that the parking is surrounded by roads in such way, that choosing a particular exit from it does not prevent a driver from leaving  in any direction he/she wants.

    Target

    Our project assumes really high range of its profitients. Although we can separate two main groups: business-side and client-side. Looking from entrepreneurs’ point of view our solution will help achieve better quantitative statistics - that means in general better customers flow and therefore bigger profits. The next aspect our system will cover are statistics, which can be used to target advertisements to a specific client. On the other hand car parks visitors will also profit from our solution. First of all - it will save lots of time - that is because of constant monitoring and guiding each car individually. Our project also puts high pressure on safety - we would like to include alerts and alarms of occuring dangers, i.a. co2 and smog alerts, traffic accidents and traffic jams. ParkMe can be used in multiple types of parkings: in shopping malls, public city parkings and in business buildings.

    Why FPGA?

    ParkMe system requires fast video processing for recognizing every car individually. Meanwhile, it requires parallel computations of routes and communication with server. We believe that using a heterogeneous computing system like Terasic FPGA boards is the best solution for this application. With Intel OpenVINO Toolkit it lets the system to achieve very high performance with good power management, scalability and ease of use. 

    2. Block Diagram

    General system schema

    There are two central hubs of our system. First of them are the GuideNests. They will consist of Terasic DA-10 NANO board with Intel OpenVINO Toolkit as a core structure. Sensors and indicators located on the parking will be connected to GPIO sockets of our board. For the cameras we will use USB sockets. Main functionality of the GuideNest is to recognize and guide cars in their workspace based on video inputs. For the video computation the GuideNest will use Intel Pretrained Model for recognition of license plates of camera facing vehicles prepared for OpenVINO Toolkit, which can be run as a heterogeneous system on CPU and FPGA. It also will be possible to connect already existing sensors to the GuideNest to support its computation for guidance. The on-board Linux will be used for communication purposes and as a host system for the peripherals.

    The second core hub of our architecture  is the server, which will operate using Elixir language, Phoenix Framework and SQL database. We will use HTTP Protocol and custom API to handle connection between the central server and GuideNests. The server will be able to operate with multiple GuideNests thanks to Elixir scalability. The data will be sent in a JSON format, which is the most popular solution for information exchange via web. The server will be accessible via web service, which will also display informations about parking archive data. The server will also maintain connection with alarms, because of safety reasons. Entrances and exits occupancy will also be monitored by the server.

    Block diagram

    Below shown block diagram is very simplified. Working on such a large system is really hard to predict. That is why we chose only the core functionalities - GuideNest, the server and their communication - to be shown on the block diagram. The server side does not include: archiving data and communicating with web service to provide clarity and simplicity.

    As mentioned before - our project will have two main centres: GuideNests and the server. 

    GuideNest firstly will try to reach the server and get data about crawling cars, their destination places, current parking road and routing tables. However the main functionality of the GuideNests will be recognizing and guiding cars individually in their workspaces and additionally store informations about them locally. The secondary function of the GuideNest will be also receiving and process data from sensors to detect any dangers in the workspace and - if any threat is detected - informing the server.

    The server will be able to maintain connection with multiple GuideNests. The main task of the server will be maintaining the communication between GuideNests to provide them: data about current parking load, routing tables, entrances and exits occupation, cars and their destinations and alarms. It will be also responsible for counting cars at entrances and exits and maintaining informations about entrances and exits occupancy - with prediction of future load based on trends from historical datas. The server responsibility will also be assigning the destination place to the entering car. The secondary, but very important, functionality of the server will be collecting security alerts from GuideNests and launching any alarms based on collected data.

    3. Intel FPGA Virtues in Your Project

    Using in our project boards supported with FPGA will have many advantages over standard CPU’s. We named some of them below:

    Performance boost - Thanks to the speed of computation done by FPGA we are able to process video from multiple cameras very fast, which lets us identify car’s license plate and key features like car’s color. The fast computation will provide smooth work for our system.

    Adapt to changes – FPGA boards are really easy to reconfigure. This feature will ensure easy and updates for our system during whole development process and eventually in future, when the product will become commercial. Moreover, the architecture of a system based on heterogeneous computation makes expanding and adapting of the ParkMe system easy and fast. Also worth mentioning is scalability of our system – FPGA supported boards with Linux operating system are really easy to get and configure. With our project we would like to provide a fully scalable solution.

    Power consumption – FPGA processors are known of their relatively low power consumption in comparing to their computing speed. In the era of global warming it is really important for us, as developers, to reduce power consumption as much as we can. We want to be sure, that our solution is eco friendly.

    4. Design Introduction

    Purpose of ParkMe!

    Main purpose of our design is to develop a solution, that will in general clean up car parks. We want to archive this through a system, that will guide each car individually from the entrance to its destination park place and all the way back to the least occupied exit. Main reason why we want to develop this project is because of time, which we are spending and wasting on underground car parks – looking for a suitable place or in traffic jams.

    Application scope & targeted users

    Our solution main application scope is to be used on underground car parks to guide cars. The system is targeted to all closed parkings: cities’ parkings, trading malls’ parkings, private underground parkings. Every parking with regulated traffic and controlled entrances can use ParkMe system. The application scope in the development phase is concentrated on developing the algorithm of the system, however, it is possible to extend the system according to the clients’ requirements even for very big infrastructures. The scalability of the system makes it possible to expand the system without reconfiguration of the software, just with adding further GuideNests and connecting them to the systems.

    Main functionalities of our system are:
    - to monitor entrances and exits,
    - to maintain a database with cars on the car park,
    - to maintain a database with archive data,
    - to guide each car individually to its destination, regarding current car park traffic and driver preferences (p.e. location of shops near lift to the parking),
    - to gather and analyze statistics of the car park.

    The simplified algorithm of the ParkMe system consists of:

    1. A driver entering a car park tells the system (p.e. by a touch screen at the barrier) where he/she wants to go,

    2. The system finds a free parking place as close to the desired building entrance as possible,

    3. The route from the entrance to the parking place is calculated and the parking place is reserved for this particular client,

    4. The driver is guided through the parking to the place chosen for him,

    5. The driver parks and goes to the desired destination,

    6. After returning to the car the system leads the driver to the least occupied exit from the car park.

    The below scheme show an example situation on a parking. The driver of the blue car, while entering the parking, choses on the destination choosing panel that he wants to go to the certain shop. There is a staircase leading to the given shop. System knows where the closes free parking place is and what is the traffic on all parts of the parking. While the car is driving down the parking, GuideNest recognize it by its license plate and shows adequate directions.

    The pool of our beneficiates will be quite large:
    - drivers – mostly because of their saved time and stress, automation of a carpark can also lead to new ways of monitoring and preventing any dangers,
    - owners of car parks – because of better flow on their possessions – better flow will result in more customers and less expenses on personal regulation of the traffic,
    - environment – less crawling on the car park means smaller CO2 emission.

    Why FPGA?

    The main reason of using FPGA supported board is to provide better system efficiency and speed. FPGA processors are known of their computing performance, which is really important in a case of our project. That’s why FPGA boards in our project will process video from cameras in our car park system. FPGA processors are also really good at supporting parallel computation. To prove that in our project we assume that each FPGA board will have to process video from not a single, but from a few cameras at once. This is necessary for the system to provide enough computational power to provide fluent guiding of the driver through the parking.

    5. Function Description

    The server will also keep statistics about the parking – load of nodes iIn our project we assumed that there will be two core hubs of our data processing: GuideNests and the global server. This description is also divided in two sections according to these core components.

    GuideNest

    Main functionality of the GuideNest is to navigate a crawling car to its destination place. To do so, it will use cameras located on key intersections and Intel pretrained models for Intel OpenVINO Toolkit to recognize a car, its license plate and key features. The next thing is exchanging data with the server to get recognized car destination. After doing so, the GuideNest will guide the current car knowing the car park layout, park places occupancy, and routes load to its destination place using the boards with light signals. GuideNests will have a possibility to use already existing park places occupancy indicators or a camera-recognition based one.

    The core assumption is that the GuideNest is connected to the server. To provide system stability and reliability, GuideNests will also be able to work with corrupted connection with the server. That’s why when the GuideNest is not connected to the server it guides locally to not occupied park places and gathers data about cars located inside of its area. When the connection is restored it will exchange the data with the server to update car locations and destinations.

    Every GuideNest has got connection to server, connection with IP cameras over ethernet and serial connection with direction indicator. The video stream from IP cameras is captured and processed by the FPGA. If there is a license plate recognized in the video, the license plate number is passed to a second program, which performs request to the server. After reciving response form the server with information about what direction should be displayed to the driver, an adequate command is send over serial connection to the direction indicator. The device highlits an appropriate direction sign for the driver.

    Entrance GuideNest

    Entrance GuideNest has got one additional feature - direction panel. The panel lets the driver to choose a destination point (e.g. certain shop, cinema, foodcourt, etc) while entering the parking. The panel can be created in any tchnology, it connects with GuideNest over serial connection. The entrance GuideNest waits for the driver to click a button. Then it checks what license plate number was read by the FPGA in video stream from IP camera, sends request to the server, informing about a new car, its license plate and destination. Server responds with the first direction for the car. The entrance barrier can be open and the direction sign can be highlited.

    The Server

    The main functionality of the server will be collecting data from GuideNests and controlling entrances to, and exits from the parking. Entering and leaving cars registration plates will be recognized and the database of the cars on the parking will be updated according to this data. The server will also process GuideNests requests – receiving the current cars’ locations and sending back their destinations. 

    The server will also be responsible for assignation of the GuideNest area to the entering car based on park places occupation gathered from GuideNests. The entering car will be associated with an ID and the driver will choose a destination place from the panel. The data will be registered in the database and the car will be ready to be guided via the GuideNests. The exiting car will be erased from the database of the cars currently located on the parking and will be added to the historical archive.

    Next functionality of the server will be monitoring occupation of exits according to the quantity of leaving cars in a period of time. This information will be used by GuideNests to navigate a leaving car to the least loaded exit.

    In association with hour, highest occupation periods daily, weekly, monthly, yearly, the least and most occupied entrances and exits. This data will be also used to predict future exit occupations and prevent them by properly leading cars. Moreover, it is possible to retrieve data about clients from the parking monitoring: occupation place (from license plate), area of interest (from desired shop area chosen at the barrier), time spent between parking and leaving. These data can be very valuable for business analytics and marketing.

    Secondary functionality of the server will be monitoring of any threats on the carpark signalized by the GuideNests – concentration of dangerous fumes, weather alerts and other troubles those can occur on the car park. IF the security systems detect any threat to the users of the parking, an adequate alarm will be started. 

    Our server will use Elixir language with Phoenix Framework and SQL database. Elixir is a dynamic, functional language designed for development of highly scalable applications. It uses Erlang Virtual Machine, which is known for running low-latency, distributed and fault-tolerant systems. In performance tests Elixir takes leading places when it comes to high concurrency and speed. That is why it makes it so perfect fit for our project. Phoenix Framework, which uses Elixir, will simplify the connection handling.

    Server implementation

    Dijkstra’s Algorithm - one of the main functionalities of the server is to guide cars which are crawling around the parking. To do so we implemented a Dijkstra’s Algorithm to calculate shortest possible path for a car to get to its destination. In the algorithm Nests are treated as nodes and average passage times between them are treated as weights. On the database there is implemented a table which contains all the neighbours and weights, which are passed to our algorithm.

    Monitoring passage times - based on the knowledge of current car location of each online car the server caclulates average passage times between Nests, when they are submitting requests to get recognized car suitable direction. Based on the knowledge where and when this car has been seen last time the server adds a record in passages datatable with calculated passage time. To manage passage times and calculate averages the server has an independent background worker, which removes old passage times from the corresponding datatable and calculates current average passage times, storing them in known neighbours datatable.

    Communication with Nests - to communicate with nodes there are implemented 3 endpoints. They are responsible for adding new cars to the online cars table, archiving exiting from parking cars and to get current direction to guide a car in on route Nests. They are implemented accordingly to RESTful API standard. Nests are submitting requests with right parameters and based on them the server knows what should be done.

    Panel for administrators - on the server there is implemented a basic panel for administrators, which shows current load, currently crawling cars and statistics about historical cars from archive. This panel can be used to provide even more statistics, but we assumed, that this is out of scope when it comes to the topic of the competition.

    Detailed description of the system usage

    1. A driver entering a car park tells the system (p.e. by a touch screen at the barrier) where he/she wants to go. He/she can choose from certain options like “food court”, “shop A”, “shop B”, ... , ”anywhere close to the entrance”. Additional button would inform the system that the driver is a disabled person to provide a route to a disabled parking place. The car’s license plate and key features are identified by the vision system.

    2. The system finds a free parking place as close to the desired building entrance as possible, considering a free parking places map and traffic map, provided by the server. 

    3. The information about the destination, connected with the car’s license plate, is stored on the server. The parking place is reserved for the particular car and cannot be reserved for another one.

    4. Route computation is done in the GuideNest and an adequate direction for the particular driver is shown on the indicator.

    5. The driver drives through the parking. Further GuideNests are recognizing the car by the license plate and key features and the adequate signs are shown on the indicators. The route calculations are based on up-to-date routing maps from the server. In case of straying from the route, a car’s destination is known by the system and it’s route can be recalculated. Every occurance of a car in certain point is posted from the GuideNest to the server with a timestamp, which lets the server to compute the traffic load.

    6. The driver reaches his park place, parks the car and goes to the desired destination. The system recognizes the car as “parked” if it enters a certain alley of the car park and does not reach next GuideNest in certain time (a bit longer than the average time of passing this alley).

    7. After returning to the car and entering the traffic the car reaches the first GuideNest from its parking place. It is recognized as “leaving” and the system guides it to the nearest exit or one of further exits if an excessive traffic occured in the nearest one. This solution is based on a presumption that the parking is surrounded by roads in such way, that choosing a particular exit from it does not prevent a driver from leaving  in any direction he/she wants.

    The part of the system described above is working as an “ad-hoc” system - GuideNests are querying the server as often as a car approaches them and the server is calculating it’s maps as often as it gets new data from the GuideNests or free parking places monitors.

    Possible further development

    The second part of the system is the analytical part. After recognizing the car’s license plate an information about it’s origin (place of registration) can be stored in the system, providing information for the car park’s owner about flow of the customers, for example: days of the week when many people from certain city come to certain shops. Such information can be useful for advertisement and organisation purposes and are very valuable for big shops or shopping malls.

    Additional part of the development is building the second type of the device, a WatchNest, which can software lets the count free parking places by the video stream from cameras in the car park. This part of the solution is less innovative, because such systems already exist in the market, therefore we decided not to focus our work on it. However, developing the WatchNest is necessary to provide a complete parking solution for the customer.

    6. Performance Parameters

    After project implementation we were able to check whether it meets our performance goals.

    In terms of system stability no problems were found. The server-side performance, due to used technologies, was high enough to provide stable and immediate responses for any requests from GuideNests. The performance of the FPGA itself was also fully satisfying – the image processing was fast enough to ensure real-time guiding and prevent CPU overloading. However, we did not manage to use the system with two cameras per one FPGA due to issue with OpenCV library. The issue, related to capturing image frames from camera, caused drastic fall of framerate after connecting the second camera: with one ip camera connected we achieved about 25-30 fps (full camera bandwidth). However, after connecting the second camera the framerate was only about 2-4fps, which is far not enough to provide stable recognition of moving cars. Moreover, connecting the second camera caused latency varying from 2 to 7 second, which is unacceptable in our solution. We did not manage to fix this issue without editing the OpenCV Libraries. This can be done in further development to give the ability to use second camera per one FPGA. That will maximize the usage of the processor and save a lot of money when it comes to use our solution in a commercial environment. For the test purposes we used only one camera, which provided us enough efficiency for real-time video processing.

    We decided to use the OpenVINO detection model provided by Intel as an example because we believe that we wouldn’t be able to create a neural network as good as the one created by Intel: we did not have access to any license plates’ photos database to provide learning resources for the neural network and, due to law regulations, we were not able to use photos of license plates taken on the streets without their owners permission. The detection and recognition model performance was good enough to provide license plates names without many errors.

    Our system is based on pre trained neural network provided by Intel. The neural network was trained on Chinese license plates, which font is different than font on Polish license plates. In many countries like China or Germany the font of the license plates is designed to be hard to mistaken – there are no two signs that could be mistaken even if viewed upside down or in reflection. For example, in Germany “6” has got a standard shape, while “9” isn’t closed totally, “O” (letter) is more round and “0” (number) is less round and isn’t totally closed. Unfortunately, the font used on Polish license plates has got some very similar signs. For example: “6” and “9” are the same character but upside down, “7” is the same character as “Z” but without the lower line. That causes a problems with license plate recognizing and we experienced it during our tests. We managed to provide real outcomes for a given video frames, what is presented in our presentation video, however the system is not perfect with license plates in Poland. We believe, that implementation of our system would be more successful in countries with better font on license plates.

    On the image above you can see the German license plates. Please note the difference of signs: 7 and Z, 0 and O, 6 and 9.


    On the image above you can see the Polish license plates. Please note the similarity of signs Z and 7, 6 and 9.

     

    When it comes to the server - performance is really great due to used technologies. PostgreSQL and Elixir combined togheter are working really well. That is why getting a direction, which a car should take to get to its destination, takes only around 20-30 ms from getting request, via calculating Dijkstra's Algorithm, to sending a response. Tests were made on the example parking map, which contained one entrance, one exit and 5 inner nodes, which is not really a big parking place, but we think, that even for bigger car parks it shouldn't take much longer.

    7. Design Architecture

    Hardware

     

    Park Me system consists of central server with database and multiple GuideNests connected to it. All GuideNests are based on CPU and FPGA heterogenous system and every one has got an IP camera connected to it over ethernet and a direction indicator, which shows the cars which way they should go. The entrance GuideNest has one additional module – direction selection panel, which is used by drivers entering the parking to chose a place where they would like to go. Direction indicator and direction selector are connected over serial connection and they are based on simple microprocessor boards. Each GuideNest is connected to the main server over ethernet connection. The cameras should be mounted on the ceiling, above the road.

    Software Flow

    Entrance GuideNest

    When a car approaches the parking entrance it has to stop by the barrier. There, the driver uses the direction choosing panel to inform the system where he wants to go. The destination panel sends information about chosen destination over serial connection to Python script running on the GuideNest device. Python script after receiving a destination reads the output from license plate detection (done simultaneously by the FPGA) and performs request to the server. In the request, formatted as json, there are 3 values send: GuideNest id (for identification purposes, which lets the parking to have more that one entrance), car license plate and its destination. The server calculates the best route to the destination and responds with the direction which should be passed to the driver. Python script receives the server response and sends request over serial connection to direction indicator to lit a certain direction arrow (or “stop” sign). Meanwhile the barrier is opened and the car can enter the parking.

    Further GuideNests

    After entering the parking a car is riding according to the direction indicators. When the car is approaching a crossing, a GuideNest placed on this crossing sees it and recognizes its license plate. Python script running on the device performs a server request as soon as a new license plate is detected. It sends GuideNest’s id (it is necessary for localizing the car on the parking map) and recognized license plate. Server calculates best way for the car to the destination choosed at the entrance and sends a response with it to the GuideNest. The device receives the server response and sends to the direction indicator command to lit a certain direction arrow.

    The Server

    There are 3 endpoints used to communicate Nests with the server:

    • /visitors/entrance - responsible for adding new cars. Entrance Nests are sending a POST request to this endpoint with a parameters containing car license plate and its destination. Server adds a car to the online cars datatable with corresponding parameters, calculates best possible route to the destination place and responses with direction, which car should take to arrive to the destination with fastest possible time.
    • /visitors/get_destination - on route Nests are asking the server with a GET request containing car license plate and their ID which way should a car take to get to its destination point. Server is calculating the best possible route with Dijkstra’s Algorithm and responding with next Nest ID and direction.
    • /visitors/exit - Exit Nests are sending DELETE request to the server, which is containing an exiting car license plate. The server moves a corresponding car from the online cars to the archived cars datable and responses with status 200.

     

    Except these 3 endpoints there are also views for the system administrators. These views are containing informations about current parking load, online cars and archived data.

    • homepage - contains basic informations about current load
    • /visitors - contains informations about currently crawling cars
    • /archived_cars - contains historical statistics about cars

    Moreover the server has also an background worker, which is responsible for calculating average passage times between Nests and removing old passage times from the corresponding datatable. Data, which is required to do so, is collected from requests mentioned before - based on them it is possible to check if the car has changed its location and how much time it did it take. This worker sets weights for Dijkstra's Algorithm, which is used via the server to calculate route for each specific car.



    3 Comments

    Aleksandr Amerikanov
    An interesting project.
    Please describe how you are going to handle the situation of careless parking when one car is parked for 2 spaces.
    Also, all the machines have different sizes, and where the smart will rifle without problems, some kind of Toyota Land Cruiser may not fit.
    🕒 Jul 06, 2019 03:29 PM
    EM022🗸
    Hello Aleksandr!
    Thank you for your time and revision of our project. We assumed that every spot on the car park, except a place for disabled people, has the same size. That is why size of the car did not bother us. Although this problem may be solved by recognizing car size on the entrance. If some kind of big arrives to the entrance - he could be guided to the wider place (there are some "family friendly" places already on big parkings). That was our first thought when we discussed your question.

    When it comes to careless parking - we assume that the parking will use a vision monitoring. Therefore, the system will be able to detect an incorrectly parked car. If such situation occur, the blocked parking places will be marked in the system as occupied and the parking's manager will be notified to inform the careless driver to move his/her car.
    However, developing the parking spots monitoring system isn't our main goal: such systems already exist and they can be developed without the FPGA-supported computation as well. Considering this, we decided to focus on the routing and guiding features.

    We are only talking about assumptions, beacause our project can be really huge and its development will be a gigantic challange. Our first and main goal is to create a guidance assuming that people are aware of how 'badly' parked car irritates others :)
    🕒 Jul 07, 2019 08:13 PM
    Bing Xia
    Hi team, please upload your project design a.s.a.p, the deadline is closing.
    🕒 Jun 28, 2019 08:17 AM

    Please login to post a comment.