Systems, Design and Development Track

In this session, Vecna Robotics’ Chief Architect, Innovation and Product Development, Siddharth Chhatpar, and CEO, Daniel Theobald, will describe a unique design method that can be summed up by the phrase – “Build to Learn.” This approach encourages engineers to start prototyping early in the design process, even before they have all the necessary information, so they can drive innovation, extract key insights into potential problems, and reduce risk. The process is unconventional and strays from industry norms, but it empowers product designers to take risks, learn from them right away, and to iterate to solve problems in real-time. In this session, Chhatpar will provide an overview of his “Build to Learn” process, offer steps to change traditional design protocol, and share ways organizations can adapt this idea into practice at their business.

Manufacturers of aerospace parts, heavy machinery, and other industrial equipment require large, automated manufacturing systems for painting, surface-finishing, inspection, and more. These systems often employ a gantry outfitted with an industrial manipulator to extend the work envelope over many meters. Due to their sheer size and associated costs, these systems must perform multiple operations to ensure adequate return on investment. Automating these large, multi-process-systems has challenges including motion planning for redundant kinematics, calibration of 3D scanning sensors, environment management, user-interface, and collision monitoring. In this session, Matthew Robinson ROS- Americas Program Manager at Southwest Research Institute, which manages the ROS-Industrial Consortium Americas, will share recent experiences with several complex automation systems for large-scale manufacturing that have recently been placed into production where ROS was used to enable and support new capabilities.

As the demand for highly complex robotics and autonomous systems applications grows, including the need to comply with industry standards such as ISO 26262 or DO-178B/C, so too does the need for high-performance, easy to use, and fully integrated development environments that can support robotics developers. Engineers must verify that the design meets requirements, is functionally correct, complies to certification standards, and is correctly implemented. Simulation with Model-Based Design (MBD) is a key capability to help understand the behavior of complex designs of robotics and autonomous systems. This talk will discuss how Model-Based Design can support modeling, simulation, verification, and in some cases automatic code generation, for robotics systems. Verification and validation techniques, required to achieve high quality and increase productivity, will also be discussed.

This presentation describes how the fusion of outdoor mobile manipulation, computer vision, and artificial intelligence, coupled with motion planning and collision avoidance algorithms is fueling the rapid adoption of supervised autonomy for outdoor applications. In this session, RE2’s Amanda Sgroi will present the key features of outdoor, mobile robotic arms, as well as how multi-modal computer vision software and artificial intelligence software can power autonomous behaviors. Likewise, PickNik’s Director of Engineering, Nathan Brooks, will present the features of the MoveIt open source motion planning and collision avoidance software. This session culminates with descriptions on how these enabling technologies, when combined, are expediting development of supervised autonomy by using construction as an example application.

High-quality data visualization is critical to robotics development and operations. Engineering teams need the ability to easily visualize everything from raw sensor data, to perception and tracking data, and planning and controls outputs. Product and Operations teams must have the ability to fetch raw, high-resolution data for anomalies in the field, and inspect these for triage and remediation. However, the current tooling landscape is fragmented – applications often do not prioritize cross-platform deployment, team collaboration, or integration with centralized data storage systems. In this session, Foxglove CEO Adrian Macneil, will share his experience developing open-source visualization stacks, and also discuss how the next generation of tooling can provide robotics companies with better access to, and analysis of, sensor data, providing a competitive advantage in the process.

A common phrase we hear stated today is that “hardware is a commodity”. In this informed session, Dave Rollinson, Chief Engineer of HEBI Robotics will discuss why this is not true when it comes to robotic actuation. He will describe what aspects of actuation are important for bringing robotic technology to new applications, particularly the need to control actuator force and the need for a flexible developer-friendly modular architecture. Rollinson will also present some recent examples of how actuation technology can be used to rapidly create and deploy robotic systems.

Any Internet search using the terms “artificial intelligence” (AI), “cognition”, or “machine learning”(ML) with the word “robotics” will return a slew of research and commercial initiatives focused on focused on robotic decision making, object identification, vision processing, autonomous navigation, motor control, sensor integration and other functions, as well as speech, facial and emotion recognition. In this panel session, attendees will learn how the latest AI and machine learning software and hardware technologies and techniques are currently being employed in ground based, aerial and maritime systems to make robots more intelligent and functional. Emerging commercial opportunities will be highlighted.