Human-agent collectives

Individuals are energetically pursuing new techniques and technologies within their specific domains of expertise. But does this add up to an organisational strategy that will help in an environment where savings are becoming harder to attain and where quality and productivity are always under pressure to improve?

Innovative mining leaders are increasingly looking to advanced technologies related to artificial intelligence (AI) for answers. Indeed, the field of AI and machine learning is beginning to reshape the mining sector in planning and operations. Maptek has become an established leader in this area.

Real success comes from more than just adopting the latest technologies. An organisational culture must interleave humans and computers in profoundly new ways.

Leading computer scientist, Nicholas Jennings has been key within the science of Human-Agent Collectives (HAC) in which humans and software agents collaborate in a seamless manner. He says that the focus to date has been on systems where all the agents are either software or hardware.

Jennings sees that ‘it is both necessary and beneficial to involve humans, working as active information gatherers and information processors, in concert with autonomous software agents’.

This has implications for mining companies with their vast array of resources. Operations need management and paradoxically management also needs the ability to run autonomously. Via HAC, mines can discover opportunities themselves, such as where the behaviour of a system depends not on its individual parts but on their relationships.

Real-time optimisation through grade control, short-term scheduling and reconciliation needs management to analyse data and react against destructive patterns and established procedures. This can lead to positive emergent ways of solving problems and developing new techniques.

Dated linear techniques for data analysis need to be replaced by agents with non-linear approaches to problem solving, using advanced deep learning techniques.

An example of enhanced human–computer learning is digital twinning where a virtual model is fed real-time data from the field. Scenarios can be quickly tested, and operations and production can be optimised. Pre-testing decisions on people and assets in a replica system leads to better outcomes.

HAC can uncover three hidden opportunities. First, a growing history allows a mine to learn about itself. Second, intelligent agents trawling through all live operational and stored historical data can look for variations. Third, collaborative negotiation can occur among and between agents and humans.

HAC promises mining companies, small and large, improvements in yields, speed and efficiency. Maptek believes HAC can make a big impact. The requirement is to:

• Understand how to provide flexible autonomy that allows agents to sometimes take autonomous actions, while at other times being guided by closer human involvement.
• Discover how groups of agents and humans can exhibit agile teaming to achieve joint goals, disbanding once the cooperative action has been successful.
• Elaborate the principles of incentive engineering in which rewards are designed so that the actions taken encourage socially desirable outcomes.
• Design an accountable information infrastructure that allows the veracity and accuracy of seamlessly blended human and agent decisions, sensor data and crowd generated content to be confirmed and audited.

These are not entirely new areas. However, the HAC context introduces additional complexity and brings new elements to the fore. According to Jennings, ‘centralised control is no longer possible. The volume, variety and pace of information and services has become too great’.

Change is inevitable. Mining companies will need to embrace this to survive in our extremely competitive environment.