This paper describes an interactive tour-guide robot, which was successfully exhibited in a Smithsonian museum. During its two weeks of operation, the robot interacted with thousands people, traversing more than 44 km at speeds up to 163 cm/sec. Our approach specifically addresses issues such as safe navigation unmodified and dynamic environments, short-term human-robot interaction. It uses learning pervasively all levels software architecture.

This paper describes a new model for understanding natural language commands given to autonomous systems that perform navigation and mobile manipulation in semi-structured environments. Previous approaches have used models with fixed structure infer the likelihood of sequence actions environment command. In contrast, our framework, called Generalized Grounding Graphs, dynamically instantiates probabilistic graphical particular command according command's hierarchical compositional semantic...

This paper describes Minerva, an interactive tour-guide robot that was successfully deployed in a Smithsonian museum. Minerva’s software is pervasively probabilistic, relying on explicit representations of uncertainty perception and control. During 2 weeks operation, the interacted with thousands people, both museum through Web, traversing more than 44 km at speeds up to 163 cm/sec unmodified

Speaking using unconstrained natural language is an intuitive and flexible way for humans to interact with robots. Understanding this kind of linguistic input challenging because diverse words phrases must be mapped into structures that the robot can understand, elements in those grounded uncertain environment. We present a system follows directions by extracting sequence spatial description clauses from then infers most probable path through environment given only information about...

This paper presents our solution for enabling a quadrotor helicopter to autonomously navigate unstructured and unknown indoor environments. We compare two sensor suites, specifically laser rangefinder stereo camera. Laser camera sensors are both well-suited recovering the helicopter's relative motion velocity. Because they use different cues from environment, each has its own set of advantages limitations that complimentary other sensor. Our eventual goal is integrate on-board single...

In this paper we describe our open-source robot control software, the Carnegie Mellon Navigation (CARMEN) Toolkit. The ultimate goals of CARMEN are to lower barrier implementing new algorithms on real and simulated robots facilitate sharing research between different institutions. order for be as inclusive various approaches possible, have chosen not adopt strict software standards, but instead focus good design practices. This outlines lessons learned in developing these

Spoken dialogue managers have benefited from using stochastic planners such as Markov Decision Processes (MDPs).However, so far, MDPs do not handle well noisy and ambiguous speech utterances.We use a Partially Observable Process (POMDP)-style approach to generate strategies by inverting the notion of state; state represents user's intentions, rather than system state.We demonstrate that under same conditions, POMDP manager makes fewer mistakes an MDP manager.Furthermore, quality recognition...

This paper describes an implemented robot system, which relies heavily on probabilistic AI techniques for acting under uncertainty. The Pearl and its predecessor Flo have been developed by a multi-disciplinary team of researchers over the past three years. goal this research is to investigate feasibility assisting elderly people with cognitive physical activity limitations through interactive robotic devices, thereby improving their quality life. robot's task involves escorting in assisted...

This paper describes a motion planning algorithm for quadrotor helicopter flying autonomously without GPS. Without accurate global positioning, the vehicle's ability to localize itself varies across environment, since different environmental features provide degrees of localization. If vehicle plans path regard how well it can along that path, runs risk becoming lost. We use Belief Roadmap (BRM) [1], an information-space extension Probabilistic algorithm, plan trajectories incorporate...

Ships often use the coasts of continents for navigation in absence better tools such as GPS, since being close to land allows sailors determine with high accuracy where they are. Similarly mobile robots, many environments global and accurate localization is not always feasible. Environments can lack features, dynamic obstacles people confuse block sensors. We demonstrate a technique generating trajectories that take into account both information content environment, density environment....

Automatically building maps from sensor data is a necessary and fundamental skill for mobile robots; as result, considerable research attention has focused on the technical challenges inherent in mapping problem. While statistical inference techniques have led to computationally efficient algorithms, next major challenge robotic automate collection process. In this paper, we address problem of how robot should plan explore an unknown environment collect order maximize accuracy resulting map....

Natural language interfaces for robot control aspire to find the best sequence of actions that reflect behavior intended by instruction. This is difficult because diversity language, variety environments, and heterogeneity tasks. Previous work has demonstrated probabilistic graphical models constructed from parse structure natural can be used identify motions most closely resemble verb phrases. Such approaches however quickly succumb computational bottlenecks imposed construction search...

Our goal is to develop models that allow a robot understand natural language instructions in the context of its world representation.Contemporary learn possible correspondences between parsed and candidate groundings include objects, regions motion constraints.However, these cannot reason about abstract concepts expressed an instruction like, "pick up middle block row five blocks".In this work, we introduce probabilistic model incorporates expressive space spatial as well notions cardinality...

This paper proposes a statistical method for calibrating the odometry of mobile robots. In contrast to previous approaches, which require explicit measurements actual motion when robot's odometry, algorithm proposed here uses sensors automatically calibrate robot as it operates. An efficient, incremental maximum likelihood enables adapt changes in its kinematics online, they occur. The appropriateness approach is demonstrated two large-scale environments, where amount odometric error reduced...

Speaking using unconstrained natural language is an intuitive and flexible way for humans to interact with robots. Understanding this kind of linguistic input challenging because diverse words phrases must be mapped into structures that the robot can understand, elements in those grounded uncertain environment. We present a system follows directions by extracting sequence spatial description clauses from then infers most probable path through environment given only information about...

Speaking using unconstrained natural language is an intuitive and flexible way for humans to interact with robots. Understanding this kind of linguistic input challenging because diverse words phrases must be mapped into structures that the robot can understand, elements in those grounded uncertain environment. We present a system follows directions by extracting sequence spatial description clauses from then infers most probable path through environment given only information about...