Connect with us


Ali Asmari, PhD, Head of AI and Machine Learning at ULC Technologies – Interview Series




Ali Asmari, PhD is head of AI and machine learning at ULC Technologies. ULC Technologies is regarded as a pioneer in robotics engineering and technology development for the energy, utility, and industrial markets. Since its start in 2001, the focus of ULC has always been the enhancement of utility operations and the support of infrastructure improvement.

What initially attracted you to robotics and AI?

I was very good at math and physics in high school which led me to study mechanical engineering in college. My favorite topics in college were Machinery Dynamics and Nonlinear Control, both of which are necessary in control of robotic systems. These topics give you all the necessary tools to turn your robotic imagination into reality. I not only built my own robots in college, but I also competed in international robotic competitions all over the world. I also decided to further study the field and entered graduate school to become a roboticist.

Machine learning was a concept that became very popular in application in early 2010. After taking a couple of basic courses in Machine Learning and Neural Network, I immediately started applying the methods in my research and work. I am personally amazed by how similar machine learning concepts are to the way that the human brain learns and functions. Use of machine learning in robotics is relatively new and has a long way to go and I feel very fortunate to be a part of this movement.

ULC Technologies has many robots that are designed to go below ground in some difficult terrain. What are some of the challenges behind building fully autonomous obstacle avoidance and path planning for these types of robots?

Much of our work has been focused on the inspection and internal remediation of older pipelines in urban areas and within these pipelines, debris is commonly found which makes fully autonomous solutions challenging. As a solution, ULC developed commercial pipeline robotic systems which are manually driven through gas mains. Over the past 20 years, our expertise in pipeline robotics has expanded which allows us to now integrate more elements of automation and machine learning.

One such initiative is called Distribution Network Information Mapping (DNIM), which is a collaborative project with UK gas network, SGN to apply machine learning to pipeline networks so that we can efficiently identify and map the pipeline and features within the pipe. This data will eventually help open pathways to obstacle avoidance and path planning in these highly complex pipeline environments.

What are some of the current robotic solutions that are offered? 

ULC works with utility and energy companies to develop and deploy robotic solutions for inspecting, repairing, and maintaining above and below ground infrastructure such as pipelines, LNG plants, substations and other complex environments.

We developed a robot called CISBOT which enters live cast iron gas mains and travels through the pipe to internally seal the joints, which helps gas networks prevent any leaks and extends the life of the pipe for up to 50 years, all without shutting off gas to customers. ULC also developed a suite of robotic camera and crawler systems for inspecting live gas mains, helping utilities reduce risk, improve efficiency, and resolve operational challenges.

Outside of our current underground robotic systems, we also have an in-house R&D team that is working on robotic solutions for other industries. One example is the Robotic Roadworks & Excavation System (RRES), a project that we are developing in collaboration with UK company, SGN. RRES is an all-electric robotic system made to replace conventional methods of excavation with capabilities that include below ground sensing, coring and cutting roadways, automated soft touch excavation, pipe installations and then the ability to reinstate the roadway. Through further development, we hope to expand the range of operations that RRES can perform in the future.

This is just a sampling of the robotics solutions that we currently offer, but additional information about our technologies can be found on our website. We have many other projects in development and are always looking to partner with companies in utility, energy and industrial industries who are looking for automated solutions.

What type of data is collected?

ULC Technologies builds custom robotic solutions to address different technical challenges in the industry. Based on the type of application, every robot captures different type of data from its environment. The following is a list of some of the popular type of data that we collect throughout our inspection operation:

  1. High Resolution Colored Images. As an example, our Unmanned Aerial Vehicles (UAV) capture 40MPixel images during inspection work
  2. 3D point clouds. An example of this is the 3D point clouds that some of our in-pipe crawler robots collect.
  3. Some of our above ground robots process LIDAR data for navigation
  4. Infrared Images. Our UAVs and Asset Identification and Mapping (AIM) solution can capture infrared images during inspection work for condition assessment of assets.
  5. High frequency radar. The RRES (Robotic Roadworks and Excavation System) uses Ground Penetrating Radar to map out location of buried assets under the ground.

There are many more different types of data that some of our platforms collect for different purposes that are not included in this list because of their specific application to one industry.

Could you discuss how these images are geotagged?

On every robotic platform, geotagging the captured images takes place unique to that system and the available information in its environment.

Our AIM system uses an onboard GPS to map out the path of our survey vehicle. Using other onboard sensors, computer vision algorithms and target tracking, our proprietary software measures the location of every identified asset and geotag their images accordingly. In GPS deprived environments such as inside an underground pipe, our robots use other methods to communicate with the above ground survey vehicles to geotag the captured data from inside the pipe.

What are some of the machine learning technologies that are currently used to process the data?

There are three main methods of machine learning that are being used in robotics and autonomous data processing, all of which are being utilized in different applications at ULC Technologies.

  1. Supervised Learning, where ground truth is necessary to train the model. These models have higher accuracy in data processing. ULC’s AIM solution is utilizing this model to identify above ground electric infrastructure assets with high accuracy and repeatability.
  2. Unsupervised Learning, in which the model identifies similarities and anomalies in the data. We have utilized this method to process the camera footage from our in-pipe crawlers and map their location along the pipe.
  3. Reinforcement Learning, which is a reward-based system to train complicated devices without complicated reverse kinematic calculations. We are utilizing this method to operate the robotic arm on the RRES platform to carry out different excavation operations.

Is there anything else that you would like to share about ULC Technologies?

We are always looking to partner with leaders in the industrial, energy and construction industries to identify and collaborate on the development of innovative solutions. Through our work and field testing, we continue to enhance our AI and machine learning capabilities and look forward to solving new challenges for our customers in the future.

Thank you for the great interview, readers who wish to learn more should visit ULC Technologies

Antoine Tardif is a Futurist who is passionate about the future of AI and robotics. He is the CEO of, and has invested in over 50 AI & blockchain projects. He is the Co-Founder of a news website focusing on digital assets, digital securities and investing. He is a founding partner of unite.AI & a member of the Forbes Technology Council.