Search

In this paper, we present an approach to the online implementation of a gait event detector based on machine learning algorithms. Gait events were detected using a uniaxial gyro that measured the foot instep angular velocity in the sagittal plane to feed a four-state left-right hidden Markov model (HMM). The short-time Viterbi algorithm was used to overcome the limitation of the standard Viterbi algorithm, which does not allow the online decoding of hidden state sequences. Supervised learning of the HMM structure and validation with the leave-one-subject-out validation method were performed using treadmill gait reference data from an optical motion capture system. The four gait events were foot strike, flat foot (FF), heel off (HO), and toe off. The accuracy ranged, on average, from 45 ms (early detection, FF) to 35 ms (late detection, HO); the latency of detection was less than 100 ms for all gait events but the HO, where the probability that it was greater than 100 ms was 25%. Overground walking tests of the HMM-based gait event detector were also successfully performed.

The vestibular system detects rotational velocity and linear acceleration of the head in order to identify body’s spatial orientation. Impairments of these organs induce the subject to spatial disorientation, postural instability, self-motion perception deficits, and chronic disequilibrium. Gaze stabilization performances are usually evaluated by the assessment of the Visual Acuity during head movements. This work introduces a new testing device, named Up-Down Chair (UDC), able to generate full body linear accelerations in the vertical direction. These stimuli can be tuned to mimic those encountered during natural activities. This platform has been designed to serve as an experimental set up to quantify vestibular functions and their relationship to clinical scales. Moreover, the UDC could be used to assess specific testing performance for a vestibular neuroprosthesis.

In this paper, we develop a system for which applications in the field of personal navigation are planned. In the current version, the system embodies a Global Positioning System (GPS) receiver and an inertial measurement unit (IMU), composed of two dual-axis accelerometers and one single-axis gyro. The IMU is positioned at a subject's foot instep, and it is intended to produce estimates of some gait parameters, including stride length, stride time, and walking speed. Data from GPS and IMU are managed by a DSP-based control box. The computations performed by the DSP processor allow to detect subsequent foot contacts by a threshold-based method applied to gyro signal, and to reconstruct the trajectory of the foot instep by numerical strapdown integration. Features of human walking dynamics are incorporated in the algorithm to enhance the estimation accuracy against errors due to sensor noise and integration drift. All computations are performed by the DSP processor in real-time conditions. The foot sensor performance is assessed during outdoor level walking trials. The traveled distance estimated by inertial dead-reckoning is compared with the estimate produced by GPS in experimental conditions where GPS can be used as a reference source for accurate absolute positioning. Results show the remarkable accuracy achieved by foot inertial sensing.

In this paper we illustrate the main features of a vision-based 2D localisation system, endowed with the real-time capability of estimating the location of active light-emitting markers disposed on a horizontal plane. The vision-based system revolves around using an inexpensive, off-the-shelf web-camera under computer control: the system is simple to install and operate, yet accurate to a level which is acceptable in a range of applications. Specifically, we describe how we intend to exploit the localisation system in an application whose ultimate goal is to support motor-impaired users in the approach to a specific configuration with their mobility aid, e.g., an electric wheelchair. To serve this application, the vision-based localisation system is integrated in a human-machine-interface, built around an handheld PC, which aims at providing the driver with a "top-view" representation of the scene. In our preliminary experiments, the perceptual enhancement is shown beneficial in increasing the docking accuracy up to few millimeters and tenths of degrees from the "target" configuration.

In this paper, we present bird's-eye (BE) viewer - a sensor-based navigation system intended to assist motor-impaired people driving their mobility aids to a specified goal. The BE-viewer is built around a vision-based 2D localization system based on a monocular image system composed of one web-camera, and endowed with real-time capability of estimating the location of active light-emitting markers. The 2D localization system is integrated in a human-machine-interface, which provides the user with a "top-view" sketch of the current and desired configurations. We introduce a protocol for the functional assessment of the docking procedure, here interpreted as a pursuit tracking task. The protocol allows to characterize how subjects use their resources to achieve a balance between speed, accuracy and control effort, depending on whether the perceptual enhancement by the BE-viewer is available or not. The benefits of the BE-view approach are presented and discussed on preliminary experiments where normal subjects perform docking procedures using a motorized anterior walker.

The problem of developing a sensing system for robots moving in a partially or completely unknown environment is investigated. A model of a generalized sensitive skin based on different types of sensors to be located on the whole surface of a robot is illustrated. A logic and hardware structure of the low-level layer of the skin is proposed to guarantee modularity and consistency of sensory data. A configuration including ultrasonic and infrared proximity sensors was chosen for the implementation of a prototype of the skin. A possible physical layout for obtaining a conformable skin to be used on any kind of robot surface is also presented. Local and general rules for low-level sensory data fusion are defined, and possible improvements of real-time obstacle avoidance path planning strategies are discussed by introducing sensor-based instinctive behaviors. A simulation of a modified potential field-based real-time planner for a mobile robot operating in a partially unknown environments is discussed to demonstrate the feasibility of this approach.

In this paper, the authors describe a sensing method, called differential compassing, which can be applied to pursue leader/follower behaviors in the field of service robotics. Differential compassing is applied to integrate proximity/range sensing in robot systems, the operation of which can involve a hands-free interaction mode, where the robot (the follower) reacts to the movements of a walking user (the leader), so as to maintain close proximity to him. Differential compassing exploits two compasses, one of them on the leader and the other one on the follower. The simple idea behind this approach is that, if a magnetic disturbance affect the two compasses, one compass will be more influenced than the other, but heading errors may have similar magnitude

The authors describe their approach to the design of redundant proximity sensor systems that are capable of extracting geometrical information about the surrounding environment and perceiving other relevant features of the reflecting objects, e.g. colour. Potential applications of the proposed sensorial systems are in the field of robotics and autonomous mobility.