Numina delivers real-time mobility insights from streets to make cities more responsive — namely, by building a street-level data utility to support walkability, bikeability, interactivity, and efficiency in cities. Our standalone sensor platform uses computer vision to measure how people and objects move throughout streets and public spaces. Our end-to-end solution delivers real-time intelligence, helping planners and municipal governments understand, map, and improve how streets and public spaces are actually used. The insights that are producible from Numina data include counts of each type of traveler or object, as well as their speeds, paths, directionality, proximities to one another, dwell times in key locations, and more — all of which accessible through a custom data dashboard and a real-time API.
Numina builds a computer vision sensor and data platform purposefully designed for transportation planning, traffic safety, and urban automation. Our proprietary sensor and software platform — made with love in Long Island City, New York — mounts to street infrastructure (such as light poles) and uses machine learning and artificial intelligence to measure the volumes, conditions, movements, and interactions of pedestrians, bicycles, different classes of vehicles (single-passenger vehicles, buses, trucks), and other objects (infrastructure, trash, unplanned obstructions) in streets, plazas, and public spaces. The insights that are producible from Numina data include counts of each type of traveler or object, as well as their speeds, paths, directionality, proximities to one another, dwell times in key locations, and more.
CHALLENGE QUESTION / OPPORTUNITY AREA:
In Miami-Dade County, Numina aims to provide multimodal traffic counts, travel paths, and safety assessments of key intersections to provide congestion data, proxies for the ridership potential of a selected transit platform, and overall help the County quantify the impact (as well as model the potential impacts) of transit on safety, environmental efficiency, and speed of travel.
This solution will make the County’s transportation system adaptable to commuter needs that may change throughout a day, week, or month — subsequently enabling the introduction of multimodal and flexible transportation options. Our continuous data streams (accessible both through our dashboard and through our API) enable our users to see how public spaces and streetscapes are actually used, inclusive of all kinds of travellers and accurate across modes. We enable planners and transportation engineers to adjust transit schedules according to traffic flow; designers can model how various design interventions will impact safety and curb usage; and rideshare, bikeshare, and electric vehicle companies can manage the distribution of their fleets based on the County’s daily (or even hourly) needs— among many other flexible use cases. In the flexibility of our sensors & data, Numina’s solution will aid the County in its responsiveness to the residents’ changing transportation needs; in the accuracy of our data across modes, Numina will aid the County in increasing the viability of the existing transportation network for non-motorists.
Due to the flexibility of its vision-based design, Numina is capable of delivering more sophisticated insights than simple object recognition. The system can analyze the relationships between multiple travelers and static objects at once objects, collecting additional metrics like average speed differentials and proximity between different modes of traffic (which, when high, pose a higher fatality risk in the event of crashes), and near misses between vehicles and non-motorists, data which is currently inaccessible to planners.
The real-time nature of Numina data also makes it possible for the system to trigger city services where needed, when needed. Numina’s real-time data could help maintenance and public safety departments identify anomalies (e.g. obstructions like trash piles or potholes, or sudden behaviors like crowds running in one direction) and automatically deploy maintenance or emergency services, or automatically provision traffic enforcement services to address idling rideshare vehicles at the curbside or in loading zones and create new revenue for municipalities.
Since the system’s information protections are consistent across any type of object detection, such enforcement can still be conducted in a private way. Numina data can also be used by advocacy and research organizations spanning other disciplines, such as economic development, social science, public health, and demography. To facilitate such collaborations, Numina has established a collaboration with Gehl to automate the collection of detailed public life data according to Jan Gehl’s methods.
PRIVACY & SECURITY:
Numina’s system was designed from the start with privacy as a foremost concern, and we take every measure to prevent the use of our technology as a surveillance tool. We minimize the data we collect and transmit, and will never incidentally or deliberately collect personally identifiable information (PII). For more information on Numina’s privacy protections and de-identification protocols, please refer to our Privacy Protocols.
No commitment to privacy can be upheld without also ensuring security. Our sensor system is designed using the defense-in-depth philosophy to ensure we deliver on our privacy commitments and protect our customers. Numina sensors process all raw data (images) it ingests onboard the device itself, only extracting key metadata from each image, and then discard the images to protect citizen privacy. The resulting data is real-time and anonymous, making Numina the only computer vision-based sensing platform purpose-built to provide street-level intelligence without surveillance. The sensors communicate with our backend over an LTE cellular network, allowing them to transmit measurements every minute and apply new capabilities on-demand via over-the-air software updates. Our hardware and software are hardened to provide a secure, robust data platform by minimizing attack surfaces and maximizing reliability. All communications are performed using strong encryption between endpoints, and all of our systems undergo periodic security review and are kept updated with the latest patches.
In addition to providing our data via our dashboard, subscription to our platform also includes access to our standard JSON API, which publishes results up to the minute and which can be easily integrated into any number of applications, data analysis, or processing tools outside of our proprietary technology stack.
The system is small and lightweight, and can be installed by one person on a ladder or in a lift truck bucket. Numina’s easy installation procedure simply requires mounting the sensor to a light or signal pole (or, if required, the exterior of buildings) using a galvanized steel strap system identical to that used for any pole-mounted sensor or street signage, configuring power to the sensor, and setting it up using a mobile installation app. The sensor can be powered using grid current wired through the pole or, alternatively, by a pole-mounted standalone solar module.
Upon installation and setup, the system requires a one-to-two week calibration period, during which Numina’s algorithms are trained to recognize objects of interest specifically against the background of each sensor’s view. Images are sampled at randomized intervals throughout the day, anonymized using Numina’s de-identification protocols (explained further in our Privacy Protocols) and transmitted from each sensor to Numina’s backend. Images are then loaded into a results verification loop, where human workers validate or correct the presence and location of each anonymous traveler or object. Numina’s classifiers are then re-trained with the new results. Numina’s bicycle and pedestrian classifiers have achieved >95% accuracy across our installed network of sensors.
Numina’s sensors work line-of-sight and provide coverage over a 90-degree field of view, with a detection range of 125 feet (approximately 38 meters) and total coverage area of 12,134.4 ft² (1,127.32 m², or 0.28 acres).
STATE OF DATA:
As of yet, no accurate data exists for understanding intermodality and multimodality on urban mobility. Planners, designers, and engineers need baseline multimodal traffic data in order to equitably plan the County’s transit to serve users from a range of destinations and utilizing a range of last-mile modes. Other infrastructure improvements to supplement a potential rapid transit route will require the same baselines. Numina is unique in its ability to demonstrate and measure the integral ties between land use and transportation, and inform strategic improvements that can result in effective mobility options in areas with auto-oriented development patterns.
Miami-Dade is in the process of planning and implementing a number of massive transportation projects (namely the Strategic Miami-Area Rapid Transit Plan, or SMART), and needs good data to streamline its efforts across modes. Much of the County’s development plans border the County’s major transportation corridors (planned or existing) and intersect surrounding mixed-use areas— meaning that the majority of the County’s plans for transportation development will have direct impact on traffic as well as retail, recreational, and residential areas. This is a positive choice and represents a critical opportunity for equitable development, but poses challenges to effectively incorporating multimodal transit access to and from stations and along the corridors after the implementation of new transit lines, and which must be planned for in a safe and efficient way.
In addition to the mixed-use nature of Miami-Dade County’s published SMART development plans, they are also large and capital-intensive projects. In order to ensure that they serve local constituents, efficiently and responsibly allocate public funds, attract downstream investment, and deliver returns on investment, the County needs data for long-range planning that holistically accounts for the impacts and utilization potential of new public transit plans and emerging transportation technology. Deploying Numina along the proposed corridors allows the County to collect this data and conduct before, during, and after assessments to concretely validate the above success metrics for development.