Algoryx

Kenneth Bodin and Hugo Börjesson represent Algoryx in the Swedish Space Delegation to India. - We are deeply impressed by Indias space program and the tremendous access to top class engineers, says Kenneth Bodin, CEO and cofounder of Algoryx.

I recommend this if you're looking for a postdoc! If you are particularly interested in the simulation and computational modeling aspects of turning heavy machinery into intelligent robots, I will soon announce open positions (phd, postdoc) at Umeå University with the possibility of collaborating with Reza Ghabcheloo's team in Tampere, Algoryx Simulation and others.

The simulation-to-reality gap is the difference between a real-life mechanical system and a digital replica of the same system fed with identical control signals. Narrowing this gap is key if you want to train an autonomous control system in a virtual environment and successfully deploy the same control system on a physical machine. In the context of earth-moving, PhD student Koji Aoshima and associate professor Martin Servin have conducted a study to examine and quantify the sim-to-real gap. Field tests with a manually controlled, real-life wheel loader digging into a pile of gravel were compared to corresponding simulation scenes created in AGX Dynamics from Algoryx. The gravel was simulated in two different ways: ▶ Using discrete element modeling (DEM) to simulate the gravel as particles. ▶ Using the multiscale terrain modeling methods of the AGX Terrain module, a heavily reduced and accelerated approximation of the full DEM model. Overall, the experiments show a 10 percent gap between the real machine and the simulations. The researchers would expect the sim-to-real gap to get even narrower with a more sophisticated model of the wheel loader for simulation. Interestingly, the simulations that used the AGX Terrain module to represent the soil dynamics delivered a level of realism that was quite comparable to the DEM-based simulations – but 100 times faster. Even in real-time, the AGX Terrain-powered simulations demonstrated a high level of realism. These experimental results highlight AGX Dynamics’ strength as a virtual training environment for autonomous machines and vehicles – especially when dynamic terrain is involved. NOTES: ▶ Koji Aoshima and Martin Servin work at the Department of Physics at Umeå University. ▶ The field tests with a real-life wheel loader were conducted by Komatsu Ltd.  LINKS: 🎓 Read the full-length paper “Examining the simulation-to-reality gap of a wheel loader digging in deformable terrain”: https://lnkd.in/dbpBeVqN (Published in Multibody System Dynamics)   🔶 Read more about AGX Terrain: https://lnkd.in/dH9JTBqH #AGXDynamics #Sim2real #DEM #simulation #autonomousvehicles #earthmoving #wheelloader #controlsystems #digitalphysics #constructionequipment #offroadautonomy #dynamicterrain

For the first time, researchers have succeeded in creating a self-driving forest machine controlled by artificial intelligence. In a research study at Umeå University, an AI system that can operate a 16-ton machine without human intervention was developed. The study has been carried out in collaboration with Skogforsk and Algoryx Simulation. “The results show that it is possible to transfer AI control to a physical forest machine after first training it in a simulated environment,” says Viktor Wiberg, researcher at Algoryx Simulation, whose doctoral thesis at Umeå University forms the basis of the work. A deep reinforcement learning controller was trained in a virtual environment using Algoryx’s physics simulation software, which accurately calculates machine dynamics and interactions with terrain and tree logs. The AI was trained in advance on Umeå University's supercomputer in several million training steps. By successfully transferring the autonomous control system from the simulated model to the real forest machine, the research study shows that bridging the sim-to-real gap for complex heavy machinery in unstructured environments is possible. As a strategy for developing autonomous control of heavy off-road vehicles, the simulation-first approach has huge advantages. “In a virtual environment, the training takes place without risk of injury and without fuel consumption,” says Martin Servin, associate professor at Umeå University and senior principal scientist at Algoryx Simulation. READ MORE ABOUT THE RESEARCH STUDY: Press release from Umeå University: https://lnkd.in/dMNyxega Research article in the June 2024 issue of the journal Robotics and Autonomous Systems journal: https://lnkd.in/dzJuYP_t #autonomouscontrol #robotics #forestry #heavyequipment #heavymachines #autonomousdriving #simulation #DRL #ROS #AGXDynamics

Developing autonomous off-road machines is not just about driving from point A to B. It often involves solving complex dynamic tasks. At Algoryx, we excel at fast and accurate simulations with realistic sensor data. Our customers turn to our simulations to speed up their development of efficient and safe autonomous control systems. Simulations are used to train, test, and validate AI systems in virtual environments, including complete scenarios. In this example scenario, a wheel loader: ☑️ Digs into the soil ☑️ Fills the bucket ☑️ Unloads the bucket into a truck The truck then: ☑️ Hauls the material ☑️ Unloads the material in a designated area The material loaded will, of course, impact the vehicle dynamics and the control of the vehicle. It affects load balance, acceleration, braking, and energy consumption. 🔸 Want to learn more about scenario simulation with AGX Dynamics from Algoryx? Contact our sales team to discover how our software solutions can propel your autonomous control efforts to success: https://lnkd.in/dSFQ4vAM

All employees are offered to work 40% of their time remotely, from home or elsewhere. Everyone is also welcome to work 100% at the office. Some employees work nearly 100% remotely on special agreements. For remote work, we ensure that the work environment is good, regarding ergonomics, air quality, noise and lighting conditions and temperature. Almost all meetings can be attended as hybrid meetings and all collaborative work is highly digitized.

Algoryx Simulation CEO Kenneth Bodin about operating in the AI space at #WASP4ALL - balancing techno-optimism, regulation-first, fundamental science, human-centered innovation, friction of bureaucracy, and massive VC reshaping the innovation landscape.

Algoryx is proud to be the main sponsor of the yearly recurring Robot Contest at Umeå universitet (Umeå University), organized by students in the MSc Engineering Physics program. The main contest is taking place today at MIT Place in the MIT building and is also being broadcasted on YouTube. The final begins at 12:30 CEST, so don't miss it! 🤖 🍿 🏅 https://lnkd.in/g5HHUJPp (Photo by Jonas Lidström from the 2023 Robot Contest finals)

🥏 Aerodynamics and collision dynamics. Office disc golf putting is all about #physics! Disc golf champion of Algoryx Simulation: Fredrik Nordfelth 🥇 #office #sports #discgolf #discgolflife #discgolfeveryday #fridayfun

Great that Sweden also is an Artemis partner! The Artemis program depends on simulation! How else can know how to build machines to explore the Moon, Mars and asteroids? Simulation of concepts, for engineering design, for early error detection and minimising design flaws, for system verification and validation, for developing and testing control systems, for training of AI systems, for astronaut training, for combinations with real sensor data in digital twins. Get in touch to see what we can do for your mission!