Gymnastics for robots

Aug 29, 2011
Not a Halloween decoration but a graphic representation of an algebraic equation that allows the calculation of spaces in which a robot cannot be controlled due to geometric reasons.

A current research project of the Austrian Science Fund FWF focuses on the motion safety and efficiency of robots. The objective of the research is to improve the control of robot movements with the help of new mathematical approaches. For this purpose, the primary focus lies on the identification of situations in which a collision may occur and the planning of an optimal motion path. Crucial factors for the initiation of this project were new developments in the areas of motion planning, computer-aided design and algebraic geometry. Within the framework of this project, algebraic methods for the control of robot motions are being used for the first time in combination with numeric and geometric methods.

A robot is born, so to speak, with motor skills. By way of program codes, motor skills are deposited into its microprocessor-based memory device. This is how motion patterns are defined, but they are not always safe for the environment – or for the robot itself. They also do not always live up to the users' expectations regarding the efficiency of the robot's performance. These limitations partly result from the that are chosen for the calculation of the motion patterns. It is now the aim of a research group led by Prof. Manfred Husty and Prof. Hans-Peter Schröcker at the Faculty of Civil Engineering of the University of Innsbruck to develop new approaches for more safety and efficiency.

Smooth Like a Robot

Within the framework of this project, which is supported by the Austrian Science Fund FWF, "robots" take on a quite general meaning. Be it an industrial robot or R2D2 – they all must first and foremost perform safely. "Safe means that a robot does not collide with its environment or with itself while performing its tasks. It also has to avoid what we call dangerous areas. This does not, however, involve physically present obstacles, but positions in the proximity of which the robot cannot be controlled due to geometric reasons, and thus may become dangerous. In robotics, we also call these areas singularity," explains project leader Prof. Schröcker.

Naturally, there are a number of methods available for calculating such areas and for avoiding them when programming motion patterns. However, according to Prof. Schröcker, all of these methods have one thing in common: "The methods that have been applied thus far are based on numeric calculations. These are not accurate. They also don't allow a guarantee that all dangerous areas have been considered in the calculations. Algebraic methods, however, are not only more accurate, but could also allow such a guarantee."

Roughly speaking, numerics deal with solving continuous mathematical problems by means of discretised computer calculations. Algebra, however, uses variables, which allows for more accurate and more comprehensive calculations. In recent years, algebraic methods have in fact become increasingly popular in Mechanical Sciences. And the first theoretical results for their application in the area of motion are also available. These have, however, not yet been adapted to the practical requirements of mechanical engineering. This is exactly what is being done in this current project.

The Journey is not the Destination

Developing new mathematical methods aims at ensuring not only that robots optimally avoid "dangerous" areas, but also that they carry out motion sequences as efficiently as possible; in this context, can mean "using the shortest route" or "quickly", "elegantly" and "in an energy-saving manner". When calculating the necessary paths, possible collisions and "dangerous" areas must be considered, as Prof. Schröcker explains: "The calculations can be thought of as connecting points in space by means of smooth curves which remain sufficiently far away from the areas that are to be avoided. Developing adequate methods for doing so is an essential objective of this FWF project that will combine fundamental mathematics and practical use."

Explore further: Robots unlikely to take big bites out of employment, expert says

Provided by Austrian Science Fund

5 /5 (1 vote)
add to favorites email to friend print save as pdf

Related Stories

Tropical Storm Dolly forms, threatens Mexico

1 hour ago

Tropical Storm Dolly formed off Mexico's northeastern coast on Tuesday and headed toward landfall in Tamaulipas state, threatening to spark floods and mudslides, forecasters said.

Halliburton pays $1.1 bn for Gulf of Mexico BP spill

1 hour ago

Oil services company Halliburton said Tuesday it would pay a $1.1 billion settlement over its role in the 2010 Gulf of Mexico oil rig blowout that led to the United States' most disastrous oil spill.

Scientists make diseased cells synthesize their own drug

1 hour ago

In a new study that could ultimately lead to many new medicines, scientists from the Florida campus of The Scripps Research Institute (TSRI) have adapted a chemical approach to turn diseased cells into unique manufacturing ...

Modern population boom traced to pre-industrial roots

1 hour ago

The foundation of the human population explosion, commonly attributed to a sudden surge in industrialization and public health during the 18th and 19th centuries, was actually laid as far back as 2,000 years ...

Recommended for you

Netflix unveils new way to share recommendations

4 hours ago

Netflix is giving its Internet video subscribers a more discreet way to recommend movies and TV shows to their Facebook friends after realizing most people don't want to share their viewing habits with large ...

Future solar panels

5 hours ago

Conventional photovoltaic technology uses large, heavy, opaque, dark silicon panels, but this could soon change. The IK4-Ikerlan research centre is working with the UPV/EHU-University of the Basque Country ...

User comments : 1

Adjust slider to filter visible comments by rank

Display comments: newest first

EWH
not rated yet Aug 29, 2011
This all sounds like a good idea, but I don't think it's that new.

Dutch researchers Karen Trovato and Leo Dorst won the '2005 Inventor of the Year' award from the New York Intellectual Property Law Association for work using a combination of Geometric Algebra and wave mechanics to plan robot motion. This is a very general method adaptable to any sort of motion or joints and complex, dynamic environments that uses analytic solutions, although it can be implemented numerically. Their work resulted in at least one patent by the Phillips company. Its roots go back nearly a quarter century: Dorst and Trovato's paper Optimal Path Planning by Cost Wave Propagation in Metric Configuration Space, was published in 1988.