Who let the dogs out? Ever since we introduced Dingo to the family last year, we’ve been extremely excited to see what sort of innovative projects this small robot would become a part of. And we admit, we’ve been a bit impatient, and so we spoke with École de technologie supérieure (ÉTS) in Montréal about what they plan with their new pack of eight Dingos so that we could share a sneak peek with you.
Introducing the Pack to the Lab
The team responsible for wrangling these new robots is the crew of the Lab INIT Robots situated in ÉTS; a university training engineers with a special focus on the industry’s needs. The Lab INIT Robots is a research and teaching group aimed at the intuitive and natural control of mobile robotic systems. They conduct research and training activities on path planning, collision avoidance, operator’s ergonomy, robot expressive motion and robot swarm’s control.
Currently, the lab’s pack of Dingos are being employed in two new courses at ÉTS: MEC744 -– Manipulateurs robotiques (Robotic Manipulators) and MEC745 -– Robotique mobile (Mobile Robotics). The teaching labs are built over two different scenarios, in order to relate the students’ work to real-world applications. The Robotic Manipulators course simulates a machine tending scenario in a manufacturing facility, integrating safety measures, risk assessment, and grasping. The Mobile Robotics course, on the other hand, simulates a search and rescue application for which the first responder needs to remotely operate a scouting robot. Both courses combine theoretical knowledge with hands-on learning and application through the teaching labs. The labs are completed in teams of three to four undergrad students with the guidance of Lab INIT Robots grad students. This cooperative environment has been vital in facilitating three key learning experiences:
- Developing familiarity with Python Notebooks and ROS.
- Applying specific complex concepts such as the Inverse Kinematics of a manipulator, Kalman filtering, etc.
- Solving a complex robotics problem using all the available tools.
Demo of the Dingo fleet in use for MEC744/MEC745
“Robotics is intrinsically highly multidisciplinary. A single academic semester to give an overview of its challenges requires a solid mobile base, like the Dingo. The out-of-the-box ROS support allows for us to integrate our code that processes complex topics such as mapping, navigation, state estimation and inverse kinematics without much overhead.”
– David St-Onge (Associate Professor and Director of Lab INIT Robots)
Accelerating Learning With ROS Support
The most common challenge that the lab faced was allowing the students to focus on the mechanical aspects of these projects, without having them invest too much energy in the programming infrastructure. The ROS learning curve can be steep and since the students are mainly from mechanical engineering, they often have limited programming skills. Thus, the team developed a full simulation to reality teaching stack, involving remote servers at ÉTS and simulation rendering in browsers and programming through Python Notebooks.
ÉTS’ Lab INIT Robots pack of Dingos (6 of 8) equipped with vision systems and Kinova Gen3 lite robotic manipulators
Through this set-up, the students are able to learn how to use ROS without the potential headaches of setting up the system. This is especially crucial in an education environment where students want to be able to jump in quickly and get a valuable hands-on experience. Robotics students can come from a variety of different backgrounds or curriculums where ROS will most likely be new to them. As lab lead David St-Onge commented: “Robotics is intrinsically highly multidisciplinary. A single academic semester to give an overview of its challenges requires a solid base. ROS allows for us to integrate in the lab’s manipulations complex topics such as mapping, navigation, state estimation and inverse kinematics without much overhead.”
So, the out-of-the-box ROS support of the Dingos formed a knowledge base that the students could leverage and quickly accelerate their learning by jumping into more advanced tasks. For example, the lab closely worked with its faculty and graduate students to build their own ros_control driver for the Dingo, so that the students could play with wheel encoders and inverse kinematic programming. By learning to drive the robot manually they would access the wheel encoders, IMU, and wheel motors to build their own drive controller. At the end of the lab, students need to complete complex missions such as building an intuitive interface for the semi-autonomous control of a scouting robot or the safe deployment of a machine tending robot in a manufacturing environment.
Bringing Affordability to the Classroom
The core deciding factor to bring in a pack of Dingos to the lab was the platform’s robustness and affordability which could be scaled to the classroom level while also sustaining years of teaching labs, especially with the robust ROS support. With a clear focus on education and on students of varying experience levels, the ÉTS team knew they wanted a solution that could get future roboticists integrated into the ROS community. Furthermore, the team wanted to inspire students by having them working on locally sourced. While its robust build is crucial in a learning environment, the lab also emphasized the platform’s ability to be easily extended with sensors and other functionalities. The team is currently working to integrate locked kill switches to enhance Dingo’s safety and protect its usage on campus.
To carry the sensor stack for each Dingo, the lab created their own turret so they could add two Intel Realsense cameras (T265 and D435i) plus Kinova Gen3 lite robotic manipulator. The height of the turret allows the incorporation of common room features in the arm’s workspace, i.e. door handles and parts laying on desks/tables. The height of the cameras also helps cover a broader area. Both the cameras and the Gen3 lite were ideal choices for the labs as they are easy to use, straightforward to implement, and most importantly, the arm was affordable enough to equip the whole classroom.
Lab INIT Robots celebrates the 2020 holiday season
A Future with Dingo
The Dingo pack is still a relatively new addition to the university’s infrastructure, but they are already ambitiously planning to add a third course to the curriculum on the challenges of mobile manipulation (control, path planning, integration, etc.). They believe the size of their robot fleet (eight platforms) will also be crucial in conducting research on human-swarms interaction and testing some of their algorithms. For now, they are still enhancing many small features of the teaching infrastructure (software and hardware), as it was deployed for the first time in the Fall (2020).
To learn more about the work that Lab INIT Robots is doing, you can visit their website here.
To learn more about what a Dingo could offer for your next project, check out our website here.