We recently spoke to ParKam about their latest project with Curtin University, Western Australia. As the biggest project in the world of bay finding for camera technology, we were curious to learn more. Nikki Eylon, Managing Director ANZ, of ParKam was on hand to tell us about how the project came about and Graham Arndt, Director of Operations and Maintenance at Curtin University was able to tell us how the project has been received from the customer-side.
Nikki, Can You Begin by Telling Us a Little Bit about ParKam?
ParKam is an image processing, deep machine learning and AI-based parking solution. Many people ask me what a deep machine learning and AI-based solution gives. The advantage of using deep machine learning and AI is that the system constantly learns from our global projects and it gets smarter and smarter over time. This allows us an accuracy level of above 99% and we can operate our system in extreme weather (conditions). We’ve operated it already in snow and desert storms and the system can function in those severe weather conditions. Our aim is to create the most efficient and accurate parking solution. We try to do that by utilizing an existing structure and existing cameras. This can even be done with old cameras.
Where hardware is needed, we can install cameras, but we prefer to use the cameras as a multi-layer device. Potentially our clients will get more than one solution per camera and they can use the parallel streaming for a few solutions. The main one is our parking solution, but the system can function as a security system and can provide other image processing-based solutions such as traffic management and so on. ParKam provides to their client a B2B to C solution. Our clients, the businesses, are provided with the BI tool, the chosen real-time availability and also alerts of parking violation both in the bay such as (parking) overtime and also outside the bay in areas such as bus stops and so on. Anything we can view from the stream we can cover and alert on, so we’re using the same cameras for alerting on violations both inside and outside the bay. We provide data to our clients in real-time, but we also show historical data and we drill down to the bay level per hour from the day we install the system. We show both occupancy level and activity level in each bay. For the end user we developed a really unique navigation system. So, the end-user navigates from home to the bay and that includes turn-by-turn in the car park and not just navigation in the car park.
Tell Us About the Curtin University Project
The project started as an RFI (Request for Information), as Curtin University wanted to find out what new, innovative solutions were available on the market and test them. We were one of ten groups which approached Curtin and from those ten groups they chose four solutions, one of which is in-ground sensors and another which was image processing based. Curtin then monitored us on how long it took us to implement the solution, what the accuracy of the solution was, what other advantages the system offered, what kind of BI tool was available, and what kind of data could be gathered from the solution.
At the end of 2018 they awarded us the contract, which is now being implemented in over 6000 bays across 29 car parks. Curtin University, like other big locations, has issues with parking and very high occupancy levels, especially during term-time when students are on-site. In some car parks, occupancy has reached 99% or even more during the day. The aim of the project is not only to teach the university which car parks get fuller, and what issues the car parks have, it is also to balance traffic between the car parks. We have a really unique navigation system that can navigate not just to the car park, but all the way to the bay, turn by turn within the car park. The idea is to balance traffic between the different car parks.
So at Curtin University, for example, we provide them with real-time occupancy levels, that will be displayed for management in the dashboard, and for the end-user on signs. This way both parties see which car parks have more availability than others. On top of this we offer the navigation system which navigates you from home, turn-by-turn all the way to the bay, taking into consideration the size of the car and how many cars are actively searching for bays in that car park. This way the driver can be navigated to the bay nearest to their destination that will still be available by the time they he reaches it, cutting the frustration.
It is also much easier to adapt software than hardware, so if for example the university decides to change add more bays to a car park or change the configuration, it is not a problem because the software will simply relearn it. With hardware solutions, they would need to be taken out and then reinstalled, causing revenue loss. Cameras are easier to scale, so if in the future there are new car parks, it is just a case of adding new cameras, or changing the settings in the car park to maintain the system.
How Has the Project Been Received by Curtin University?
The project is one of the biggest that the university has ever undertaken, and they took advantage of the parking project to add things for other projects, such as lighting poles that were missing and things like that. And whilst it was one of the largest projects that they have implemented, everyone participating was on board with it, so it was nice to see people from all different departments excited about the project.
What’s Next for ParKam?
So ParKam is now expanding and doing more and more deployments. We are about to deploy our new product Swift Drop-Off, for airport parking. With this product, not only are we monitoring the availability of the spaces, but displaying them on large digital signs at the entrance to the drop off zone, and moving the traffic forward through visualisation of the drop-off space.
ParKam is also deploying car parks through Tel Aviv city, stage by stage. And there are many car parks globally. We are happy to see our product spreading and gaining more and more positive reaction from our customers and clients who love the solution.
Could You Please Introduce Yourself?
My name is Graham Arndt, I work at Curtin University, Perth, Australia. I am the Director of Operations and Maintenance, and my portfolio includes parking.
How Did You Choose ParKam for the Project?
The aim of the project was to improve the experience for our drivers on the campus. We began by running a think tank with academics and general staff to find out what annoyed them most and what we could do to improve their experience. After that we contacted ten suppliers who could provide part or all of the solution, and we offered to do a proof of concept with each of those. Out of the ten we ended up with four companies participating and from those we chose ParKam.
We compared technologies, such as loop counting, but this was a system we had in place prior to the proof of concept, and we knew that wasn’t accurate. We tried in-ground sensors and we also ran video analytics using number plate recognition, so it would read the number plates of cars going in and out. We then ran video analytics of the bays which can tell you whether a bay is available or not. We compared the advantages and disadvantages of each of those technologies and found that the video analytics of the bays was by far the superior of the technologies.
How Has the Improved Parking Facility Been Received?
We are still implementing the solution, but during the proof of concept we trialled guiding people to a bay. One of the main complaints, was that when drivers arrived in the morning they couldn’t find a bay where they wanted to find a bay. Parking guidance stops people driving around and about, in and out, looking over and over again at the same occupied bays. Now we can direct drivers straight to a bay which is less frustrating for them, and it is also better for the environment because cars do not need to drive around looking.
Has the Solution Worked in the Way You Anticipated?
Yes, it is very accurate and it comes with an app that directs drivers turn by turn to a bay. We found that once a driver parked in a bay, within less than two minutes it would show as occupied. It was able to look at a car park and see that there were three bays available, but also that there were three cars circulating that car park, so it would know that there would be nothing available in that car park. So it is a very artificially intelligent system.
What Are the Benefits of Using Video Analytics Compared to In-Ground Sensors?
In-ground sensors can only give information about the bay where they are located, but cameras can give information about the entire car park. So for example, our cameras can tell us if people have parked where they shouldn’t have parked, for example at the end of the bay where they can obstruct traffic, or along a bank where it is dangerous to park. So the cameras can alert us to people parking in the wrong spot, as well as telling you that people are parking in the right spot. In-ground sensors can’t do that.
You can also use the system to see if someone has been parked in a bay for too long. So if you have a 30 minute bay the cameras can learn when the car goes in, time it, and then send you an alert if the person has overstayed. But they can also tell if the person reverses out and then drives straight back into the same bay. Whereas the in-ground sensors can’t tell you that, they can only tell that the bay has been vacant for a short time. The cameras can’t be tricked easily.
When Will the Project Be Completed?
We expect to have all of the cameras up and running by the end of the month, and then the cameras will take a little time to learn what is and isn’t a car and whether it is parked in the bay correctly or not. So the accuracy begins quite low but rapidly builds up to about 98% or more as we found in the trial. And so this should be done by the end of August.
The big thing about Curtin is that we have 6000 bays that need to be monitored, so it is a lot. And of course with a camera you can monitor a lot more bays. And so the price per bay made the cameras a better option for us. I believe that this is one of the biggest way-finding projects to date in Australia.
Watch our In The Spotlight video here:
ParKam was formed to resolve the worldwide problem of finding reliable parking on the go. It is known that 30% of inner-city traffic is created by people who are looking for available parking spots.
On the other side of the spectrum, parking lots lose customers and money as finding parking is still an exhausting and frustrating process. In addition, parking lots are losing money as they lack the data to utilize maximum capacity and rely on unnecessary manpower
Our unique, innovative solution is based on Real Time Computer vision. To find the best available parking spot we use our innovative proprietary algorithms which process footage from cameras that are already installed or will be installed by ParKam where necessary. Unlike other solutions, we only need one simple camera for up to 100 parking spots.
From the same streaming, we also operate a state of the art enforcement system.
ParKam offers a holistic, fully automated, very accurate, patented solution, second to none, for your smart city, parking lots of shopping centers, stadiums, airports, universities etc.
ParKam’s state of the art navigation system guarantees you always arrive at an available parking spot.