Abstract
Formulating problems rigorously in terms of optimization principles has become a dominating approach in the fields of machine learning and computer vision. However, the systems described in these fields are in some respects different to integrated, modular, and embodied systems, such as the ones we aim to build in robotics. While representing systems via optimality principles is a powerful approach, relying on it as the sole approach to robotics raises substantial challenges. In this article, we take this as a starting point to discuss which ways of representing problems should be best-suited for robotics. We argue that an adequate choice of system representation—e.g. via optimization principles—must allow us to reflect the structure of the problem domain. We discuss system design principles, such as modularity, redundancy, stability, and dynamic processes, and the degree to which they are compatible with the optimization stance or instead point to alternative paradigms in robotics research. This discussion, we hope, will bring attention to this important and often ignored system-level issue in the context of robotics research.
Similar content being viewed by others
Notes
See [11] for a non-technical introduction.
Decentralized partially observed Markov decision processes.
Translating from microscopic behavior (described in terms of optimality principles) to the “effective” optimality principles that describes the macroscopic behavior is one of the most interesting and fundamental problems in science per se. Renormalization, cooper pairs, superfluidity, and similar macroscopic theories are examples of this endeavor—or of circumventing this endeavor [10].
Of course, we could define objective functions over a solution path as well; now desribing desirable optimization processes in terms of optimality principles.
It should be noted that for the optimization approach the result is always the end point of the generated trajectory, while for the process approach either the end point or the trajectory itself can be the result.
The “generative models” in statistics focus on the parametric specification of a probability density and do not particularly emphasize any process aspect.
References
Arkin RC (1998) Behavior-based robotics. MIT press, Cambridge
Braitenberg V (1986) Vehicles: experiments in synthetic psychology. MIT press, Cambridge
Cohen P (1995) Empirical methods for artificial intelligence. MIT press, Cambridge
Feynman RP, Leighton RB, Sands ML (1963) The Feynman Lectures on Physics, volume II. Addison Wesley, Boston
Goldman CV, Zilberstein S (2004) Decentralized control of cooperative systems: categorization and complexity analysis. J Artif Intell Res (JAIR) 22:143–174
Heskes T (1999) Energy functions for self-organizing maps. In: Oja E, Kaski S (eds) Kohonen maps, pp 303–316. Elsevier
Kearns M, Littman ML, Singh S (2001) Graphical models for game theory. In: Proceedings of the Seventeenth conference on Uncertainty in artificial intelligence, pp 253–260. Morgan Kaufmann Publishers Inc
Kohonen T (1982) Self-organized formation of topologically correct feature maps. Biol Cybern 43(1):59–69
Koval MC, Pollard NS, Srinivasa SS (2014) Pre-and post-contact policy decomposition for planar contact manipulation under uncertainty. RSS, Berkeley
Laughlin RB (2006) A different universe: Reinventing physics from the bottom down. Basic Books
Laumond J-P, Mansard N, Lasserre JB (2015) Optimization as motion selection principle in robot action. Commun ACM 58(5):64–74
Moore J, Cory R, Tedrake R (2014) Robust post-stall perching with a simple fixed-wing glider using LQR-trees. Bioinspiration Biom 9(2):025013
Mordatch I, Popović Z, Todorov E (2012) Contact-invariant optimization for hand manipulation. In: Proceedings of the ACM SIGGRAPH/Eurographics symposium on computer animation, pp 137–144. Eurographics Association
Murata T (1989) Petri nets: properties, analysis and applications. Proc IEEE 77(4):541–580
Pfeifer R, Gomez G (2009) Morphological computationconnecting brain, body, and environment. In: Sendhoff B, Körner E, Sporns O, Ritter H, Doya K (eds) Creating brain-like intelligence. Springer, Berlin Heidelberg, pp 66–83
Posa M, Cantu C, Tedrake R (2014) A direct method for trajectory optimization of rigid bodies through contact. Int J Robot Res 33(1):69–81. ISSN 0278–3649, pp 1741–3176, doi:10.1177/0278364913506757
Ratliff N, Zucker M, Bagnell JA, Srinivasa S (2009) CHOMP: Gradient optimization techniques for efficient motion planning. In: Robotics and Automation, 2009. ICRA’09. IEEE International Conference on, pp 489–494. IEEE
Rendell P (2011) A universal turing machine in conway’s game of life. In: High Performance Computing and Simulation (HPCS), 2011 International Conference on, pp 764–772. IEEE
Schaal S, Peters J, Nakanishi J, Ijspeert A (2005) Learning movement primitives. In: In Robotics Research. The Eleventh International Symposium pp 561–572
Seok S, Wang A, Chuah MY, Otten D, Lang J, Kim S (2013) Design principles for highly efficient quadrupeds and implementation on the mit cheetah robot. In: Robotics and Automation (ICRA), 2013 IEEE International Conference on, pp 3307–3312. IEEE
Thomas U, Hirzinger G, Rumpe B, Schulze C, Wortmann A (2013) A new skill based robot programming language using UML/P statecharts. In: In Proc. IEEE Int. Conf. Robotics and Automation (ICRA), pp 461–466
Toussaint M (2009) Probabilistic inference as a model of planned behavior. Künstliche Intell 3/09:23–29
Toussaint M (2014) Newton methods for k-order markov constrained motion problems. arXiv preprint arXiv:1407.0414
Traub JF, Woźniakowski H (1982) Complexity of linear programming. Oper Res Lett 1(2):59–62
Weston J, Ratle F, Collobert R (2008) Deep learning via semi-supervised embedding. In: Proc. of the 25th Int. Conf. on Machine Learning (ICML 2008)
Zarubin D, Pokorny FT, Toussaint M, Kragic D (2013) Caging complex objects with geodesic balls. In: Proc. of the Int. Conf. on Intelligent Robots and Systems (IROS 2013)
Author information
Authors and Affiliations
Corresponding author
Additional information
We thank the German Research Foundation for the creation of the Priority Programme SPP 1527, by which this research is supported.
Rights and permissions
About this article
Cite this article
Toussaint, M., Ritter, H. & Brock, O. The Optimization Route to Robotics—and Alternatives. Künstl Intell 29, 379–388 (2015). https://doi.org/10.1007/s13218-015-0379-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13218-015-0379-7