Among humans, teaching various tasks is a complex process which relies on multiple means for interaction and learning, both the part of teacher learner. Used together, these modalities lead to effective learning approaches, respectively. In robotics domain, task has been mostly addressed by using only one or very few interactions. this paper we present an approach robots that key features general people use when each other: first give demonstration, then allow learner refine acquired...

Understanding intent is an important aspect of communication among people and essential component the human cognitive system. This capability particularly relevant for situations that involve collaboration agents or detection can pose a threat. In this paper, we propose approach allows robot to detect intentions others based on experience acquired through its own sensory-motor capabilities, then using while taking perspective agent whose should be recognized. Our method uses novel...

We focus on a robotic domain in which human acts both as teacher and collaborator to mobile robot. First, we present an approach that allows robot learn task representations from its own experiences of interacting with human. While most approaches learning demonstration have focused acquiring policies (i.e., collections reactive rules), demonstrate mechanism constructs high-level based the robot's underlying capabilities. Next, describe generalization framework allow interact humans order...

Behavior-based systems (BBS) have been effective in a variety of applications, but due to their limited use representation they not applied much more complex problems, such as ones involving temporal sequences, or hierarchical task representations. This paper presents an approach implementing these AI-level concepts into BBS, without compromising BBS' key properties. We describe Hierarchical Abstract Behavior Architecture that allows for the and execution complex, sequential, hierarchically...

Hand-based verification is a key biometric technology with wide range of potential applications both in industry and government. The focus this work on improving the efficiency, accuracy, robustness hand-based verification. In particular, we propose using high-order Zernike moments to represent hand geometry, avoiding more difficult prone errors process hand-landmark extraction (e.g., finding finger joints). proposed system operates 2D silhouette images acquired by placing planar lighting...

Video surveillance has significant application prospects such as security, law enforcement, and traffic monitoring. Visual using computer vision techniques can be non-invasive, cost effective, automated. Detecting recognizing the objects in a video is an important part of many systems which help tracking detected gathering information. In case surveillance, vehicle detection classification it control statistics that used intelligent transportation systems. Vehicle poses difficult problem...

One of the foundations social interaction among humans is ability to correctly identify interactions and infer intentions others. To build robots that reliably function in human world, we must develop models can use mimic intent recognition skills found humans. We propose a framework uses contextual information form object affordances state improve performance an underlying system. This system represents objects their using directed graph automatically extracted from large corpus natural...

Horizon line is a promising visual cue which can be exploited for robot localization or geo-localization. Prominent approaches to horizon detection rely on edge as pre-processing step inherently non-stable approach due parameter choices and underlying assumptions. We present novel uses machine learning Dynamic Programming (DP) extract the from classification map instead of an map. The key idea assigning score each pixel, interpreted likelihood pixel belonging line, representing multi-stage...

Behavior-based systems (BBS) have been effective in a variety of applications, but due to their limited use representation they not applied much more complex problems, such as ones involving temporal sequences, or hierarchical task representations. This paper presents an approach implementing these AI-level concepts into BBS, without compromising BBS' key properties. We describe Hierarchical Abstract Behavior Architecture that allows for the and execution complex, sequential, hierarchically...

In this paper we address the problem of teaching robots to perform various tasks. We present a behavior-based approach that extends capabilities robots, allowing them learn representations complex tasks from their own experiences interacting with human, and use acquired knowledge teach other in turn. A learner robot follows human or teacher maps its observations environment internal behaviors, building at run-time representation experienced task form behavior network. To enable this,...

Most approaches for motion analysis and interpretation rely on restrictive parametric models involve iterative methods which depend heavily initial conditions are subject to instability. Further difficulties encountered in image regions where is not smooth-typically around boundaries. This work addresses the problem of visual by formulating it as an inference layers from a noisy possibly sparse point set 4D space. The core method based layered representation data voting scheme affinity...

As service robots become increasingly common in society, so too will accidents involving robots. Current law functions effectively to adjudicate the disputes that arise from such accidents, but as technology improves and robot autonomy grows, it much harder apply currently existing laws. Instead, new legal frameworks have be developed address questions of liability human–robot interaction. We already proposed framework 'Robots Animals', which are analogized domesticated animals for purposes...

The Technical Committee (TC) on Human - Robot Interaction (HRI) is 12 years old. next ICRA in St. Paul, Minnesota, United States, 14-18 May 2012 will be the time for TC triennial review. We propose that notwithstanding many results achieved during past years, HRI continue to play an important role coordinating activities related HRI, consolidating communities, and pushing research toward new underexplored areas. envision several possible directions future research: 1) developing smart...

For socially assistive robots (SAR)to be accepted into complex and stochastic human environments, it is important to account for subtle social norms. In this paper, we propose a novel approach socially-aware navigation (SAN)which garnered an immense interest in the Human-Robot Interaction (HRI)community. We use multi-objective optimization tool called Pareto Concavity Elimination Transformation (PaC-cET)to capture non-linear behavior, contribution community. A candidate point on trajectory...

Mass segmentation is one of the fundamental tasks used when identifying breast cancer due to comprehensive information it provides, including location, size, and border masses. Despite significant improvement in performance task, certain properties data, such as pixel class imbalance diverse appearance sizes masses, remain challenging. Recently, there has been a surge articles proposing address through formulation loss function. While demonstrating an enhancement performance, they mostly...

Feature selection is crucial in many machine learning applications as it helps to avoid overfitting, improve accuracy, provide faster training, and better interpretability. The Shapley value game theory provides a fair efficient way find the contribution of each feature prediction model. In this paper, we propose method select top features based on Advanced feature. We have used two datasets several methods 70% highest contributing features. Then dataset with only input four models. Finally,...

Producing an accurate motion flow field is very difficult at boundaries. We present a noniterative approach for segmentation from image motion, based on two voting processes, in different dimensional spaces. By expressing the layers as surfaces 4D (four-dimensional) space, process first used to enforce smoothness of and determine estimation pixel velocities, regions The boundary then combined with intensity information original images order locally define tensor field. correct inferred by 2D...