Maram Sakr | University of British Columbia (original) (raw)

Papers by Maram Sakr

Much of the morbidity and disability associated with industrial work settings arise from accident... more Much of the morbidity and disability associated with industrial work settings arise from accidents involving humans and robots. Force Myography (FMG) is a potential technique to be used as an additional control measure for safer human-robot interaction without the need for robot hardware modification or replacement. The FMG signals represent the volumetric changes in the forearm due to muscle contraction, which were acquired using a Force Sensitive Resistor strap. A 1DOF torque sensor was used to model the point of interaction between a robot and a human. The following isolated upper extremity movements were considered: forearm pronation-supination, wrist flexion-extension and wrist radial-ulnar deviation. Torque regression models based on FMG data were created with two machine learning methods: Support Vector Machine (SVM) and Artificial Neural Network (ANN). Performance indices were defined and used for the comparative study between the two learning methods. The results demonstrated the feasibility of using FMG to estimate torque with accuracies around 90%. Both methods also demonstrated strong intra-and inter-participant consistency of FMG signals. The results will be beneficial for measuring the contact force between human and robot during their interaction.

In this letter, we propose a training approach combining handover hand and trial and error traini... more In this letter, we propose a training approach combining handover hand and trial and error training approaches and we evaluate its effectiveness for both robotic and standard laparo-scopic surgical training. The proposed approach makes use of the data of an expert collected while using the da Vinci Surgical System. We present our data collection system and how we use it in the proposed training approach. We conduct two user studies (N = 21 for each) to evaluate the effectiveness of this approach. Our results show that subjects trained using this combined approach can better balance the speed and accuracy of their task execution compared with others trained using only one of either handover hand or trial and error training approaches. Moreover, this combined approach leads to the best performance when it comes to the transferability of the acquired skills when testing on another task. We show that the results of the two studies are consistent with an established model in the literature for motor skill learning. Moreover, our results show for the first time the feasibility of using a surgical robot and data collected from it as a training platform for conventional laparoscopic surgery without robotic assistance. Index Terms-Medical robots and systems, surgical sobotics: laparoscopy, surgical training, training by demonstration.

Human-machine-interfaces (HMI) have a key role for translating human intention into control comma... more Human-machine-interfaces (HMI) have a key role for translating human intention into control commands to external devices. Different wearable techniques, including surface electromyography (sEMG), have been proposed for acquiring bio-signals that reveal the human intention. In this paper, we explore an easy-to-use wearable sensor device that can be used to measure force-myography (FMG) signals. We assess if FMG signals can be used to estimate isometric torque of hand pronation-supination, wrist flexion-extension or wrist radial-ulnar using a regression model. Results of our investigation report an average accuracy over 90%. The related standard deviation of 0.02 is showing consistency of data among different data collecting sessions. The proposed FMG-based device shows therefore promising performance for different future applications, which may include: monitoring the progress of patients during exercising in arm rehabilitation programs; proportional control of robotic hand prosthesis; and control of robot movements.

Force Myography (FMG) is novel method of tracking functional motor activity using volumetric chan... more Force Myography (FMG) is novel method of tracking functional motor activity using volumetric changes associated with muscle function. With comparable accuracy and multiple advantages over traditional methods of functional motor activity tracking, FMG has made leaps and bounds in terms of applications in human-machine interfaces and healthcare devices. As a field that is rapidly gaining popularity in health innovation, the aim of this paper is to contribute to our understanding of the nature FMG methods and establish it as a robust and reliable technique. The main point of exploration for this study is the impact of sensor placement and spatial coverage on FMG methods. Five participants were invited to perform a series of isolated wrist motions and hand gestures while wearing custom built FMG devices. Linear Discriminant Analysis (LDA) machine learning models were developed using 80% of the data for training and 20% for testing. Overall, the accuracy of the LDA models ranged from 75% to 100% across all subjects and dimensions of FMG data. The model accuracy improved when increasing the spatial coverage from 1 FMG band to 2, but it did not increase further with additions. The results also showed that the improved accuracy offered by a large spatial coverage of FMG data can be approximated by lower spatial coverage if sensors were place in an optimal location. This location was indicated to be midway between the wrist and the collective muscle bellies of intrinsic forearm muscles. The knowledge generated from this work aims serve as a guide towards the development of portable FMG based technology for widespread deployment in the general population. The hope is that the long-term benefits of continued FMG research will address issues in healthcare associated with disparities in access to medical technologies.

Force Myography (FMG) is a method of tracking functional motor activity using volumetric changes ... more Force Myography (FMG) is a method of tracking functional motor activity using volumetric changes associated with muscle function. With comparable accuracy and multiple advantages over traditional methods of functional motor activity tracking, FMG has shown a promising potential in terms of applications in human-machine interfaces, tele-operation and healthcare devices. This paper provides a study that explores the effect of the spatial coverage and placement of the Force Myography (FMG) measurements on the accuracy and predictability of the machine learning models of isometric hand force. Five participants were recruited in this study and were asked to exert isometric force along three perpendicular axes while wearing custom built FMG devices. During the tests, the isometric force was measured using a 6 degree-of-freedom (DOF) load cell whereas the FMG signals were recorded using a total number of 60 FSRs, which were embedded into four bands worn on the arm. General Regression Neural Network (GRNN) model was employed in this study for predicting the hand force in three axes from the recorded FMG signals. The regression model was trained using all possible band combinations to find the optimal placement for the FMG measurements. The results showed that the accuracy significantly improved when increasing the spatial coverage from 1 FMG band to 2 or 3 bands for all axes. While the accuracy slightly improved when the 4 bands used instead of 3. Specifically, the average R 2 across all subjects and axes are 0.68 ± 0.12, 0.84 ± 0.04, 0.91 ± 0.02 and 0.95 ± 0.01 using single, double, triple and four bands combination, respectively, in 5-fold cross-validation evaluation. The knowledge generated from this work aims serve as a guide towards the development of portable FMG based technology for widespread deployment in the general population.

Currently available interfaces for programming industrial robots, e.g., teach pendants and comput... more Currently available interfaces for programming industrial robots, e.g., teach pendants and computer consoles, are often unintuitive, resulting in a slow and tedious process for teaching robot tasks. Kinesthetic teaching, i.e., teaching robot motions by placing the robot in a gravity compensated state and then moving the robot though the desired motions, provides an alternative for small robots for which safe interaction can be guaranteed. However for many large industrial robots physical interaction is not an option. Emerging augmented reality technology offers an alternative interface with the potential to make robotic programming faster, safer, and more intuitive. The use of augmented reality admits the presentation of large amounts of rich, visual, in-situ information. However, it may also overload the user's visual information capacity, or may not provide sufficient feedback regarding the state of the robot. With the addition of gestural control and tactile feedback to augmented reality, we propose a system that allows users to program and execute robot tasks in an efficient and intuitive manner, by providing relevant feedback through different channels to maximize clear communication of the task commands and outcomes.

We propose a modification of a well-known antinspired trail-following algorithm to reduce congest... more We propose a modification of a well-known antinspired trail-following algorithm to reduce congestion in multi-robot systems. Our method results in robots moving in multiple lanes towards their goal location. Our algorithm is inspired by the idea of building multiple-lane highways to mitigate traffic congestion in traffic engineering. We consider the resource transportation task where autonomous robots repeatedly transport goods between a food source and a nest in an initially unknown environment. To evaluate our algorithm, we perform simulation experiments in several environments with and without obstacles. Compared with the baseline SO-LOST algorithm, we find that our modified method increases the system throughput by up to 3.9 times by supporting a larger productive robot population.

Force Myography (FMG) is a technique involving the use of force sensors on the surface of the lim... more Force Myography (FMG) is a technique involving the use of force sensors on the surface of the limb to detect the volumetric changes in the underlying musculotendinous complex. This paper investigates the feasibility of employing force-sensing resistors (FSRs) worn on the arm that measure the FMG signals for force prediction in dynamic conditions. The predicted force value can be mapped into velocity value to control a linear actuator to track hand movements. Two FMG bands were donned on the participant wrist and forearm muscle belly to measure the FMG signals during force exertion. An accurate force transducer was used for labeling the FMG signals by measuring the exerted force. Three regression algorithms including kernel ridge regression (KRR), support vector regression (SVR), and general regression neural network (GRNN), were used in this study for predicting hand force using the FMG signals. The data were collected for 200 seconds for training the regression model. Then, the trained model was used for online force prediction for 430 seconds. The testing accuracy was 0.92, 0.90 and 0.79, using KRR, SVR and GRNN, respectively. These results will be beneficial for monitoring hand force during human-robot interaction or controlling the robot movement.

Force Myography (FMG) quantifies the volumetric changes in a limb occurring with muscle contracti... more Force Myography (FMG) quantifies the volumetric changes in a limb occurring with muscle contraction and can potentially be used to design convenient, low-cost interfaces to assist in activities of daily living (ADL). The aim of this study is to evaluate whether elders can effectively use an FMG-based wrist band to interact with their environment. In this regard, an FMG band consisted of an array of force-sensing resistors (FSRs) was designed. Ten participants were grouped in two classes, namely "senior" and "non-senior", and were instructed to perform control gestures and unconstrained ADL tasks while wearing the designed wrist band. To evaluate the usability of the band, correct identification of hand gestures and reaction times were noted. Results showed that seniors were capable of successfully performing a control gesture within 1.4 s of displaying the instruction during online testing. The individually-trained gesture identification algorithm achieved an accuracy of 76.5% in this case. Non-seniors had a reaction time of 0.9 s with an overall classification accuracy of 91.2%. This preliminary study demonstrates the potential and feasibility of utilizing FMG-based technology to provide elders with assistance during activities of daily living.