California startup VIMOC Technologies has launched Landscape-Computing to provide smart cities with a scalable API, sensor-data-driven platform. Currently testing the platform in Palo Alto, Calif., and Newcastle, Australia, Landscape-Computing uses a modular design and edge processing technologies to avoid overloading bandwidth with the weight of sensor data that, until now, has been a significant barrier to the design and deployment of smart cities applications.
In a move that echoes Elon Musk’s opening up of the source code for the Tesla vehicle to help grow the electric car industry, VIMOC has open sourced the sensor hardware technology so that cities, technology manufacturers and potential startups can build new sensor-based hardware products that can work with the API platform.
Open Source Sensor Hardware for Smart City Infrastructure Innovation
When opening Tesla’s vehicle patents, CEO Musk noted, “It is impossible for Tesla to build electric cars fast enough to address the carbon crisis. By the same token, it means the market is enormous.” VIMOC CEO Tarik Hammadou and Aaron Hector, VP of engineering and software development, share a similar mindset to Musk.
“Smart cities” is a concept that recognizes that cities can be more efficient, make better use of resources, heighten livability and better engage citizens by making use of sensors and open data to manage and deliver government services. For the idea of “smart cities” to really take off, however, there first needs to be a rapid expansion in the availability of the sensor hardware that is able to track and better measure resource usage and community needs. Hammadou and Hector have open sourced the sensor hardware they are using on the Landscape-Computing computing platform so that sensor device manufacturing is not a bottleneck for the rollout of smart city infrastructure.
“We believe we need to open source the hardware in order to push innovation,” says Hammadou. “In our experience in dealing with sensory devices, in order to reduce the cost of sensors and push innovation, open sourcing the hardware is essential.”
Landscape-Computing’s approach to building smart city infrastructure is to create a network of sensors across the city. A mesh network then picks up all the data from the sensors. This data is then processed at the edge, via the open-sourced neural box (“nBox”) technology. From there, nBoxes transmit key data to the cloud so it can be accessed and integrated by the Landscape-Computing API platform as needed.
The pipeline for the nBox goes from the input phase, acquisition, info processing, data classification for machine learning and so on. The entire system is designed in a modular fashion to allow additional functionality. It is a modular architecture; you can add more sensor types.
Sensor data gets collected on the nBox, so all the data is handled and processed at the edge. What goes through the access gateway is the state transitions (for example, with the parking sensors, we pass on the data about whether a parking space is occupied or not), all the other data gets stored at the edge key. We are capturing all diagnostic information, as well signal strength and so on, but we are only uploading the raw state transitions. This significantly reduces data flow. We are reducing to the bare amount the information that goes to the cloud, so we are talking kilobytes per day going to the cloud versus the kilobytes of data we are collecting per second.
“We have a RESTful API, so it is quite standards compliant,” says Hector. “We have already created a sample app using the API. We did it as a proof of concept but got it up and running fairly quickly. We are happy to offer support to third-party developers who want to create additional applications. At the moment, we have the access to our data there. You can download the API key and gain access to the data and start testing straight away.”
The Landscape-Computing API provides functionality for accessing data related to a variety of sensors (at the moment, this is predominantly the parking sensors that VIMOC has in place for its pilot projects) and for performing queries on aggregated information about logical groups of sensors. The REST API is built on Apache tools, and supports queries in TEXT/XML, APPLICATION/XML and APPLICATION/JSON formats.
So far, VIMOC are making the data from its two pilot projects available via the Landscape-Computing platform.
Pilot Projects: Palo Alto and Newcastle
In pilot project trials in Palo Alto and Newcastle (just north of Sydney), VIMOC has placed wireless vehicle parking-detection sensors to better manage on-street parking. Sensors can collect data such as location of parking space, when it is occupied or vacant and for how long it is occupied.
For the Newcastle study, this is also being extended with sensor data collection on pedestrian movements. The goal in Newcastle is to collect data that can help local businesses better understand how shoppers interact with retail areas in order to help revitalization and economic development strategies.
Unlike more invasive and expensive surveillance techniques, privacy and confidentiality are maintained with the use of sensor tracking; no individual data is able to be recorded. This is not about identifying when someone should be fined for staying too long in a parking space, but instead about measuring city engagement around retail hubs and the related parking needs. However, the downside to this passive sensor data collection is that the fact the data even exists can remain hidden, with residents unaware that there is the opportunity to mine and use this data themselves to advocate for things like improved public transport access.
VIMOC sees the potential for a future in which cities like Newcastle will be able to make the data and API available via their open data platforms. At present, if citizens or hackers want to make use of the data to create new civic tech, they can do so by registering for an API key, but the initial intent of the pilot is that the data will be used by the city and the local business development association to understand shopper flow through the city’s retail hub.
Hammadou shares about the pilot experience so far:
The interesting thing for us is that it has been an extraordinary learning process. When we decided to inject the same concept from Newcastle into Palo Alto, it was a completely different urban environment. There is a cultural impact of putting something like this in two different cities. When you build technology in a lab in front of your computer, it is an easy process. But building something for the Internet of Things, you need the cooperation of users and to learn from their interactions. We saw different interactions from cities and business operators in each of the two pilot cities.
Hector explains the setup of the pilots:
It was architecture first: the Landscape-Computing with the edge processing. That has really paid dividends for us, and that has really impressed the cities as we have been able to demonstrate a robust and easily expandable infrastructure. The pilot has worked really well for us because we have the technical underpinnings. It’s been a very unique way to bring hardware and software technologies together.
The approach is already gaining interest among leading smart cities and IoT forums. Cisco singled out the approach as a top-placed submission among 812 submissions to its IoT Global Challenge, while Hammadou has been invited to speak on "injecting intelligence into the DNA of the city" at the upcoming Smart Cities Expo in Barcelona, Spain (to be held directly after the first International Predictive APIs and Apps Conference, which also includes discussions of using data to help cities improve service delivery and resource use in diverse areas, from bike rental schemes to waste management).
Developers can review API documentation, test the API live in a sandbox environment and register for an API key at VIMOC Technologies' Landscape-Computing website.