We are building a decentralized protocol called CIRCUM, that incentivizes owners of geospatial sensors to pool their data. The goal is to create a large-scale, near-real-time data source for 3D maps.
The owners of geospatial sensors, which can include autonomous vehicle operators, telecom companies or surveying companies already collect this data. Additionally, other entities such as delivery vehicle operators, taxi drivers, and drone operators could collect this data if they had a financial incentive to do so. The protocol provides this incentive through a token that distributes value between data consumers and resource providers.
The collected data consists of point clouds, obtained using LiDAR sensors or photogrammetric processing. This data can be reconstructed and rendered in any 3D engine. The resulting 3D map has various applications, including urban planification and smart city, construction, telecoms, autonomous mobility, simulation, gaming and other consumer use cases.
Local governments and companies struggle with outdated and incomplete 3D map data because the acquisition process is very expensive.
Local governments fund large-scale aerial acquisition programs for tens of millions of dollars and the data gets obsolete very fast. For example, the french government has launched a 5-years long 3D modeling program of the whole country for 60M€ (LiDAR HD program). By the time the acquisition is complete, part of the 3D model will already be 5 years old. It’s too old to be efficiently used for urban planification and management of land use cases (monitoring of soil artificialization, deforestation, flooding, project insertion, heat island…).
A lot of companies use 3D maps/models for building games, improving user experience of their app, planning an event or construction work, simulating the effect of real world phenomenon (digital twinning) or supporting autonomous vehicles. These companies often have to collect their own data or rely on solutions like Google Maps or Open Street Map. Unfortunately the these solutions run into the same issues of obsolescence. Google Maps also uses low-flying aircraft for its data acquisition which is very costly. Their data is routinely +8 years old outside of the main urban centers. It is also not very detailed and looks blocky, which is not suitable for designing immersive interfaces.
Our CEO encountered these issues while working at her first company and trying to implement a 3D map into a mobile application. We believe that in 2024, this implementation should give much better results and that is the origin of Extra Labs.
In the early 2000s, Google established a monopoly in the field of 2D maps, as they were the only entity capable of making massive investments to collect large-scale data and build processing capabilities. This monopoly, combined with their substantial investments, inevitably resulted in increased prices, such as the 1600% price hike of Google Maps API in 2018.
Today, the race for 3D maps has begun, but the landscape has changed. Sensors are now widely distributed worldwide, and a vast amount of data is already collected. This means that there is no longer a need for massive data collection investments. Additionally, the emergence of distributed ledger technology like blockchain enables the aggregation of latent capacities of these sensors through tokenization, facilitating their collaboration. These factors create a situation where 3D maps can be produced and maintained at a low cost. Furthermore, this approach allows for significant improvements in level of detail and recency by directly connecting to on-the-field sensors and leveraging the “constantly expanding” feature of permissionless systems.
The company Extra S.A.S. was founded by Charlie Durand, Laurent Caraffa, and Dhruv Malik.
Charlie Durand is a repeat entrepreneur with a focus on sales and product.
Laurent Caraffa holds a PhD in Computer Vision and is a researcher on 3D surface reconstruction at the French Mapping Institut (IGN).
Dhruv Malik is a senior cybersecurity and cryptography engineer with experience in web3.
Using blockchain to coordinate resource providers and data consumers enables us to ensure a smooth service while avoiding high CapEx. The token incentivizes owners of geospatial sensors to share their data, even when there are no immediate data consumers. This helps CIRCUM build a dense and coherent capacity to fulfill requests on-demand. Additionally, by linking the token value to the actual usage of the protocol, we avoid the need to invest in purchasing data without certainty about their resale value.
CIRCUM serves as a catalyst for pooling different resources such as geospatial data, computing power, storage capacity, and 3D surface reconstruction algorithms. These resources are essential for the protocol to fulfill its mission of generating detailed and up-to-date data for 3D maps.
The protocol consists of various mechanisms that attract resources, validate them, process them, and enable easy access and consumption of data for the end-user.
To facilitate the explanation of CIRCUM, we have divided it into four layers based on the four roles mentioned above:
The first (as in deepest) layer ensures the fair distribution of the value paid by the data consumers to the resource providers.
First, each resource provider is rewarded with tokens in proportion to their work. These tokens serve as keys that grant access to the data produced by CIRCUM. They are fungible, meaning that each token has the same value as any other, regardless of the actions that led to its mining.
To access the data, data consumers need to purchase tokens from the resource providers. Once tokens have granted access to the data, they are burned, ensuring a sustainable circulating supply.
The second layer is also the most theoretical one atm. As a permissionless system, CIRCUM cannot rely on the honesty of its geospatial data providers to maintain an accurate map. These providers may share false, outdated, or low-quality data for various reasons. To address this issue, we aim to align the economic interests of all participants in ensuring the trustworthiness of the data.
To achieve this, we propose using a Proof-of-Stake mechanism combined with a Reputation Score. Contributors are required to stake tokens before they can start submitting data. If their data is frequently contradicted by other contributors or shows others signs of low quality, their Reputation Score will decrease. Once the score falls below a certain threshold, their stake will be slashed, and they will be unable to submit new data or receive any tokens until it is renewed.
This "negative" mechanism is complemented by a "positive" one that rewards the contributor as long as their data remains the best version available. Periodically, the map will be recomputed to incorporate new contributions. During this recomputing phase, new information may replace the information previously provided by a contributor if the protocol determines it to be superior (more up-to-date, better supported, more detailed). As long as the contributor's information persists into the next period (i.e., as long as the protocol doesn’t find a better version), the contributor receives tokens at each period.
This layer is especially challenging and will require extensive research. In the meantime, trust can be ensured through a partially manual validation process.
The surface reconstruction layer is responsible for storing the raw geospatial data provided by contributors and processing it. The algorithms, compute power, and storage in this layer are rewarded by tokens in the same way as geospatial data. Extra Labs should be the primary provider of algorithms in this layer and earn tokens for our contributions. The processing tasks in this layer include cropping, cleaning, and indexing datasets, computing orientation, and running reconstruction through geometric or deep learning algorithms.
This final layer is the closest to the data consumers as it manages their interaction with CIRCUM. The goal is to provide the developers with an experience similar to integrating the familiar Google Maps API/SDK. To achieve this, we will use account abstraction, allowing data consumers to access the data without directly interacting with the blockchain. However, the main challenge lies in providing the consumer with a stable data price, while the token price is not stable.
To address this, we will link the number of data requests a consumer can make ("maploads") to the price paid in fiat for a token, rather than the token itself. For example, if 1000 maploads cost $10, a token purchased for $5 will grant 500 maploads. If the price increases later on, a token purchased for $15 would grant 1500 maploads.
This approach is transparent for the data consumer, who only needs to complete a KYC process and provide their credit card information. By ensuring the token price can rise and fall following usage, we protect the protocol's resiliency. We express this solution through the equation: 1000 maploads = $10 ≠ 1 token.
We are about to complete the development of a first prototype of our protocol, which will be tested in the coming months in a experiment with the Ile-de-France region (around Paris). More precisely we have deployed the compute-over-data infrastructure that will support the protocol by enabling storage and processing of the data, and started deploying geospatial processing pipelines on this infrastructure. We have also deployed smart contracts to reward data contribution with an account abstraction mechanism.
Our company aims to develop a "lab" business model, popular in the DeFi ecosystem with companies like Morpho Labs or Mangrove DAO. It involves generating revenue by contributing to the technical development and maintenance of a public good. CIRCUM is designed to support this business model by utilizing a token that effectively distributes a portion of the protocol's revenue to resource providers.
In our case, the company Extra would be one of the resource providers for the independent protocol CIRCUM. We would earn tokens by completing tasks that require trust building or surface reconstruction algorithms. Like others resources providers, we would then sell these tokens on the market.
CIRCUM has the potential to revolutionize the funding, collection, and sharing of geospatial data. By providing a detailed, dense, up-to-date, and cost-effective data source for 3D maps, we can bring our industries into the realm of 3D and make a significant impact on urban planning, mobility, and consumer applications.
However, the impact of CIRCUM could be even greater in developing parts of the world. This includes 85% of the global population and most of the fastest-growing cities, such as Delhi, Shanghai, Dhaka, Kinshasa, Lagos, and Cairo. These cities face rapid development but struggle to provide their residents with basic services like housing, transportation, water access, electricity, and waste management. In these areas, the traditional method of collecting large-scale LiDAR data every few years is highly impractical due to the rapid obsolescence of data and the limited investment capacity of local governments.
A new funding and collection model like CIRCUM's could bring significant improvements in how local governments, NGOs, and companies understand and monitor these cities. This could be a game-changer for urban planning and urban services, as well as for other sectors of the economy, like e-commerce. Indeed, accurate 3D maps have the potential to solve the last-mile problem in a world where 4 billion people lack addresses.
Beyond 3D cartography, Circum introduces a new approach by using decentralized technologies to facilitate efficient collaboration between humans and machines in creating a public good.This permissionless method of distributing tasks, ensuring honesty, and rewarding contributions allows for the production of a more efficient and scalable service than what could be achieved in a centralized manner.
While CIRCUM has been initially designed for the use case of 3D maps, its infrastructure could actually be utilized for a wide range of other purposes. In the future, we plan to fork it to collect various types of data for digital twin purposes, including indoor and underground mapping, pollution, temperature, and more. Our work will be fully open source, aiming to inspire other projects in this direction.