The field of robotics has seen many advancements in robot-robot interaction over the last few years. Many of these have been due to artificial intelligence (AI) algorithms that allow multi agent systems to collaborate and learn. Now, with the increased interest in blockchain integration, we see this being employed alongside AI in the field of multi-agent robotics.
Image Credit: NicoElNino/Shutterstock.com
Ferrer et al. (2021) have created a system that allows multi-agent systems to communicate and follow a set of leaders successfully, despite there being malicious actors within the system. This is achieved by using a blockchain for message passing. As a result, this highlights a possible method for ensuring that multi-robot systems cannot be completely drawn from their objectives.
Byzantine Generals Problem
The theoretical underpinnings of blockchain technology come from a problem known as the Byzantine Fault.
Here, an emperor must pass a message to each of his generals to make a particular move; each general has a free, independent choice, but they must all agree; otherwise, the army will be destroyed.
From this analogy, it is clear to see that the likelihood of this occurring (all generals agreeing upon the same move) decreases as the number of generals increases.
Now, comparing the generals to the nodes of a computer network, the same situation arises – no individual computer would know what messages are sent between the others.
Blockchain provides a method to solve this problem. As a blockchain is an immutable store of data between different individuals in a network, this is an immutable store of data.
It is also highly robust against the Byzantine Generals problem as blockchains typically employ a proof of consensus algorithm to ensure that the information within the blockchain is correct and does not include adversarial attacks by malicious actors.
Common consensus algorithms implemented on current blockchains include Proof of Work (PoW) and Proof of Stake (PoS).
Blockchain and Multirobot Systems
New research has presented a multi-robot system consisting of leader, follower and maliciously acting robots (Byzantine robots), which can successfully complete an objective despite hindrances.
The system works with one or more leader robots, sensing its environment before deciding where to move. This leader then signals to move all of the other (follower) robots in the same direction.
Each of the leaders in the group will be signaling the same direction to their followers. However, within the group, there will be a number of Byzantine robots that are able to act of their own accord. This unruly action comes either in the form of using inappropriate operations while running its movement algorithm or by taking no actions at all.
These have the effect of simulating malicious robots which deviate from the group's objective, whether intentional or by mistake. These Byzantine robots can also be leader robots, meaning that it is possible for byzantine robots can send erroneous movement signals to their followers.
In the system that was designed, the identities of each robot are placed into the starting block of the blockchain and each leader receives a fixed number of tokens that are used to add transactions to the chain.
These transactions are for sending movement instructions, and one token is needed to add a transaction. By checking on the blockchain what the majority of leader robots signaled at that particular time point, the follower robots can determine if the leader sent a false message. This leader loses a token and once a robot is out of tokens, it can no longer send messages.
The system was tested by simulating several different follow-the-leader situations where the number of malicious robots was known or unknown. The researchers found that, even when Byzantine leaders initially misled follower robots, the token system enabled all of the follower robots to eventually reach the target destination.
Also, due to each leader starting with the same number of tokens, the researchers were able to determine the maximum amount of malicious directions a Byzantine leader can give.
Where Can this be Useful?
This type of technology is extremely useful in scenarios where there may be numerous robots in the same place, such as during multi-drone flights.
The work conducted here could create new security systems for robots using transaction-based interactions. This is as the algorithms enable a system designer to determine the amount of memory required to store the blockchain based on the number of robots and possible malicious actors.
The technology also helps multi-agent system designers determine the number of robots that may be needed for a particular mission and the likelihood of completing an objective, even if a fraction of leader robots act maliciously.
Continue reading: How can AI Prevent Fraud?
References and Further Reading
Ferrer, E.C., Jiménez, E., Lopez-Presa, J.L. and Martín-Rueda, J., (2021) Following Leaders in Byzantine Multirobot Systems by Using Blockchain Technology. IEEE Transactions on Robotics. Available at: https://ieeexplore.ieee.org/document/9555625
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.