Research

• Scientific challenges: Stable operation in unstructured environments, Operator ergonomics & cognitive load reduction, Social behaviour & interaction, Adjustable autonomy / shared control
• Applications: Tele-robotics, Haptics, Cognitive and Physical Interaction Systems, Assisted living, Healthcare, Rehabilitation, Medical systems
• Links: DRI health, creative industry, Medical Delta (Erasmus MC & LUMC), Space (ESA Tele-robotics & Haptics Laboratory)
• Faculties: 3ME, IO, EWI, TBM, BK, L&R

Within the TU Delft Robotics Institute, it is our goal to develop robotic systems and user interfaces that optimize and simplify the way in which humans interact with robots on physical and cognitive levels.

Applications demanding for robots interacting with humans, demand safe and robust control. Human-robot systems must be user centred, especially for non-trained users in multiple application domains. Examples can be found in health-care robots, surgery systems, household companion robots, remote manipulation robots and interactive environments in architecture.

Physical, as well as cognitive, human-robot interaction is still a relatively young field of research. Our research focus is on improving classical mono-modal interactions (tactile, force-feedback, visual, logic), and on optimally exploiting shared control concepts between a robot and the human. Cognitive interaction must be intuitive to allow for planning and predicting the actions of the robot. Moreover, physical interaction with force and tactile feedback needs to be transparent and robust over varying and unknown environments and in some instances over long time delay (e.g. between space and ground of from a ship to an underwater system).

• Scientific challenges: Human instruction interpretation, shared autonomy & safe operation, advanced handling and manipulation, world interpretation, context awareness
• Applications: Agricultural industry, public buildings (hospitals), logistics
• Links: Medical Delta, Creative Industry, Greenpoort
• Faculties: 3ME, EWI, IO, BK, TNW

We want to develop flexible robots that can do the repetitive labour in small-scale production companies. The jobs, currently done by human workers, consist for instance of picking the products from a table and putting them correctly oriented in a packaging machine (on a conveyor belt, usually). Typically, each day brings a different batch, but all are related to a small set of products and a small set of packaging actions.

The future scenario is that one employee can manage an entire packaging line consisting of, for instance, four robots that do pick-and-place motions. The single employee responsible for this packaging line can set the robots up within 30 minutes, instructing each robot what to do and what to look for. From then on his job is to supervise the line, intervene when errors occur, and perform incidental tasks such as bringing new pallets of product.

Within the TU Delft Robotics Institute, we can develop the next generation of industrial robots for small and medium enterprises. These robots are expected to be flexibly deployable, to be mobile, to receive work instructions from non-specialist employees, and to operate safely side-by-side with human employees. These robots will greatly increase the productivity of SME’s, and will prevent further production loss to lower wage countries.

• Scientific challenges: Collective system behaviour, Distributed sensor systems & networks, Miniaturization, System safety & robustness
• Applications: Traffic and pollution observation, Ship & logistic tracking, Space exploration, Self-deploying sensor networks
• Links: Delfly, Delfi n3Xt, Olfar, DRIs: infrastructure, climate & energy, Delft Institute for Research on ICT
• Faculties: L&R, EWI, TBM, 3ME
• Theme leaders: Guido de Croon and Chris Verhoeven

Inexpensive, highly miniaturized autonomously flying aircraft and spacecraft and roving robots have the capability to cooperate in swarms that can quickly provide information on large area infrastructures in harsh environments where other data collection systems are hard to deploy or use. Unmanned Aerial Vehicles (UAV’s) are already an option when ground accessibility is restricted, or simply as a cost-effective means to provide observation capabilities in places where it is temporarily needed. Examples include observation of crowds at large events, traffic observation, geological exploration, climate observation or disaster relief. Operation in a swarm adds the two key factors robustness and omnipresence and autonomous operation makes that the swarm goes where the data is. Much research is done on large sensor networks and their benefit is clear. Flexibility and cost for deployment of these networks is still an issue. Swarms are “self-deploying” flexible and robust sensor networks.

Local observations of UAV’s are easily combined with the global observations by nano-satellite swarms. This provides high resolution details where necessary in a global data set. A multi-tiered system is created that can respond flexibly to provide observation at differing levels of detail.

TU Delft has established itself as one of the leading creators of swarm elements by flying the nano-satellite Delfi-C3, launched in April 2008 and still fully functional, and by demonstrating the practical implementation of the highly-miniaturized UAV, the “Delfly micro”, carrying a camera and acoustic sensors. Both systems fully exploit the excellent soft and hardware capabilities that are unique for TU Delft. Along with the research to build these individual sensor-carrying automata, an increasingly clear picture has been established of what these highly miniaturized systems are and how they can be optimally used.

Within the TU Delft Robotics Institute, it is our first goal to create highly miniaturized spacecraft and aircraft, increase their level of autonomy and demonstrate the proper operation of these swarm elements both in the air and in space. The goal is to make them cooperate in a swarm and demonstrate the feasibility of data-collecting swarms.