VICKSBURG, Miss. – Researchers at the U.S. Army Engineer Research and Development Center (ERDC) are accustomed to finding unique and innovative approaches to solving today’s most challenging engineering problems. The current COVID-19 environment presented new obstacles for the Sensor Integration Branch (SIB) in the ERDC’s Information Technology Laboratory (ITL) and Environmental Processes Branch (CNE) in the ERDC’s Construction Engineering Research Laboratory, prompting the robotics group to capitalize on simulation software capabilities to meet mission requirements.
The Robotic Assessment of Closure Gates for Safe Entry is an SIB project spanning across several ERDC laboratories and devised to explore semi-autonomous inspection systems for use in examining gates at waterworks with earthen dams. The project, fostered under the Flood and Coastal Systems Research and Development Program, is being developed by the ITL with collaboration from the ERDC’s Coastal and Hydraulics Laboratory (CHL).
“It was one of the needs identified by the U.S. Army Corps of Engineers (USACE) Civil Works Research Area Review Groups a couple years back,” said Dr. Anton Netchaev, a research computer scientist with ITL.
Currently, these types of USACE gate inspections are performed by dam safety personnel. “Sometimes it’s dangerous to go in there,” said Netchaev. “Sometimes the conditions are not good enough to do a proper inspection.”
Natchaev explained the inherent risk associated with the inspections. “Inspections are done by humans using just cameras and ladders,” he continued. “Think about going into a large dam on a boat with a couple of people. You bring a ladder and have to walk up a long conduit — sometimes 2000 feet — up to a gate that most of the time is spraying water at high velocities, then you have to climb up on top of the gate — sometimes 20 feet high — to take pictures.”
The Robotic for Engineer Operations (REO) is another project spanning multiple ERDC laboratories designed to provide teleoperated engineer support capabilities to operate in undefined and uncontrolled environments at beyond-visual-line-of-sight standoff distances. It is part of the Next Generation Combat Vehicle Cross Functional Team and funded through Army Futures Command 6.2 and 6.3 Research Development Test & Evaluation funds. REO will deliver increased survivability of engineer teams by removing engineers from high-risk operations and providing expanded capacity and capability to support Army modernization efforts.
The team is utilizing ATV-sized amphibious robotic platforms outfitted with a variety of cameras and sensors running on the Robot Operating System (ROS). The current system uses cameras with Light Detection and Ranging, or LiDAR, data collection and wheel encoders to track robot movements and gather temporal-spatial data for assembly into an accurate 3D map of the environment.
“Our task was to design a system that is able to swim up to the conduit and then drive through it to do a preliminary inspection to see if it’s safe to send in inspection personnel. We don’t want to cut out any personnel; they’ve known the dams for a very long time,” said Netchaev. “Similarly, one of the tasks for REO is to create sufficiently detailed maps of the operational environment to enable unmanned engineer operations.”
Two years into the project, the team had successfully navigated the robot around indoor/outdoor test facilities on the ERDC campuses in Vicksburg, Mississippi, and Champaign, Illinois. The team was preparing to travel to a dam and military installation for field testing when the COVID-19 pandemic brought everything to a screeching halt.
“We were just getting to the point where hands-on development would be integral to the workflow,” said Netchaev. “It is crucial to calibrate the sensors to make sure the algorithms work. We test it, tweak some parameters, and then test it again — multiple times per day we’re trying to run this robot.”
With most teams working from home with limited access to resources and travel, researchers at the ERDC began searching for new and innovative ways to continue their work despite the current conditions. The team focused on various types of computer simulation software.
One of the simulation approaches utilized prerecorded files containing sensor data and other information on the robots’ movements. The files can be replayed repeatedly to simulate the data collection process without needing to return to the physical environment, allowing for development of data processing techniques, such as mapping and localization, to continue without ever needing to make physical contact with the robot or a dam.
“A lot of the fine tuning that typically takes place directly on the hardware is now being implemented in modeling and simulation software before we even get to the robot,” said Netchaev. “It’s been extremely useful.”
Another simulation approach entailed developing a full virtual environment for control and simulation of the robot using a Gazebo simulator, an open-source robotic simulation software that integrates seamlessly with the ROS. Through collaboration with the University of California San Diego (UCSD), the team was able to create a realistic simulation of a dam conduit for testing mapping tools against the significant localization challenges posed by a long, extremely regular conduit.
“One of the challenges that we have on this project is to localize inside a tunnel that is devoid of GPS or any kind of marking that we usually use terrestrially” said Netchaev. “The UCSD team made highly accurate maps of the tunnel that we are now using as a virtual test ground for our robot."
The collaboration between the SIB, CNE, CHL, the ERDC’s Geotechnical and Structures Laboratory and UCSD continues to prove successful, thanks to simulation efforts.
“The ultimate test will be once we put this developed system inside a real operational environment,” said Netchaev.
The robotics team anticipates taking their robotic system to various sites to test their simulated capabilities in real-world environments in the near future. Conversely, the data acquired in these field tests will be utilized for development of more realistic simulation environments for future efforts.
“What we learned from this is that we can adapt to these complex or complicated positions and still achieve our goals as long as we come together,” said Netchaev. “To learn to overcome adversity in these situations is how we continue to support our mission”