Artificial intelligence (AI) algorithms enable computers to predict certain outcomes based on large amounts of data. In construction, AI may be used to predict the risk of project cost overruns, the risk of on-site accidents, and the need for maintenance over time.
Artificial intelligence (AI) is a widely used and vaguely defined term describing when machines are able to perform actions that normally require human intelligence. As technologies evolve, the boundary between what is and is not considered to be AI keeps moving. Current examples of AI include operating autonomous cars, understanding human speech (natural language processing) and competing against human champions in difficult strategic games such as chess and Go.
Machine learning is a particular kind of artificial intelligence, describing when the machine is able to learn from experience (e.g. historic data or through tests) without being explicitly programmed. A classic example of machine learning is to teach a machine to differentiate between pictures of dogs and cats by feeding the machine with thousands of examples of each. When the machine learning algorithms has been trained it is able to classify new pictures as either cat or dog, based on its experience.
Deep learning is a particular kind of machine learning that is inspired by the way our brain works, and is based on the concept of artificial neural networks. Deep learning is heavily dependent of a large amount of data and large computing power.
AI and machine learning algorithms is behind many of the technologies mentioned in these Technology Cards (e.g. autonomous construction vehicles and generative design). Here we explicitly describe AI that is used predictions. In construction, this may include:
- predicting cost overruns based on experience from previous projects, and factors such as project size, contract type and the competence level of project managers
- predicting the risk of future accidents on site. AI can be used to analyze photos and video from jobsites and scan them for safety hazards (e.g. workers not wearing helmets).
- predicting maintenance needs and preventing unplanned downtime based on sensor data.
Benefits and challenges
- Improving the reliability of early phase cost and time estimates
- Improving safety on site through risk predictions
- Optimising the utilization of assets through maintenance predictions
- Challenging to collect the big amount of data needed for training machine learning algoritms
Application examples
The construction machinery producer Caterpillar has partnered with the technology company Uptake to develop predictive maintenance solutions for Caterpillar’s products. These solutions allow customers to monitor machinery in the field, and receive predictive maintenance advice, rather than waiting for products to break down on site (www.chicagobusiness.com).
The U.S. company Suffolk construction uses AI to detect safety hazards on site. They trained the AI on more than 700.000 pictures from 360 previous construction projects and trained the algorithm to recognize whether the construction workers were wearing hard hats, safety vests, googles and gloves. The company aims to compute risk ratings for projects so that safety briefings can be held when a threat is detected (www.autodesk.com).
Development stage
The technology has been demonstrated in the construction sector, but is still not fully implemented.
Construction impact
Artificial Intelligence predictions may affect all phases of a construction project.
Read more
www.expertsystem.com
www.datasciencecentral.com
constructible.trimble.com
www.mckinsey.com