J. B. Gafford, F. Doshi-Velez, R. J. Wood, and C. J. Walsh, “
Machine learning approaches to environmental disturbance rejection in multi-axis optoelectronic force sensors,”
Sensors and Actuators A: Physical, vol. 248, pp. 78 - 87, 2016.
Publisher's VersionAbstractLight-Intensity Modulated (LIM) force sensors are seeing increasing interest in the field of surgical robotics and flexible systems in particular. However, such sensing modalities are notoriously susceptible to ambient effects such as temperature and environmental irradiance which can register as false force readings. We explore machine learning techniques to dynamically compensate for environmental biases that plague multi-axis optoelectronic force sensors. In this work, we fabricate a multisensor: three-axis LIM force sensor with integrated temperature and ambient irradiance sensing manufactured via a monolithic, origami-inspired fabrication process called printed-circuit MEMS. We explore machine learning regression techniques to compensate for temperature and ambient light sensitivity using on-board environmental sensor data. We compare batch-based ridge regression, kernelized regression and support vector techniques to baseline ordinary least-squares estimates to show that on-board environmental monitoring can substantially improve sensor force tracking performance and output stability under variable lighting and large (>100C) thermal gradients. By augmenting the least-squares estimate with nonlinear functions describing both environmental disturbances and cross-axis coupling effects, we can reduce the error in Fx, Fy and Fz by 10%, 33%, and 73%, respectively. We assess viability of each algorithm tested in terms of both prediction accuracy and computational overhead, and analyze kernel-based regression for prediction in the context of online force feedback and haptics applications in surgical robotics. Finally, we suggest future work for fast approximation and prediction using stochastic, sparse kernel techniques.
PDF S. Russo, T. Ranzani, C. J. Walsh, and R. J. Wood, “
A soft pop-up proprioceptive actuator for minimally invasive surgery,”
The 9th Hamlyn Symposium on Medical Robotics. pp. [Best Poster Award], 2016.
PDF J. B. Gafford, R. J. Wood, and C. J. Walsh, “
Self-Assembling, Low-Cost, and Modular mm-Scale Force Sensor,”
IEEE Sensors Journal, vol. 16, no. 1, pp. 69-76. [Cover Article], 2016.
Publisher's VersionAbstractThe innovation in surgical robotics has seen a shift toward flexible systems that can access remote locations inside the body. However, a general reliance on the conventional fabrication techniques ultimately limits the complexity and the sophistication of the distal implementations of such systems, and poses a barrier to further innovation and widespread adoption. In this paper, we present a novel, self-assembling force sensor manufactured using a composite lamination fabrication process, wherein linkages pre-machined in the laminate provide the required degrees-of-freedom and fold patterns to facilitate self-assembly. Using the purely 2-D fabrication techniques, the energy contained within a planar elastic biasing element directly integrated into the laminate is released post-fabrication, allowing the sensor to self-assemble into its final 3-D shape. The sensors are batch-fabricated, further driving down the production costs. The transduction mechanism relies on the principle of light intensity modulation, which allows the sensor to detect axial forces with millinewton-level resolution. The geometry of the sensor was selected based on the size constraints inherent in minimally invasive surgery, as well as with a specific focus on optimizing the sensor's linearity. The sensor is unique from the fiber-based force sensors in that the emitter and the detector are encapsulated within the sensor itself. The bare sensor operates over a force range of 0-200 mN, with a sensitivity of 5 V/N and a resolution of 0.8 mN. The experimental results show that the sensor's stiffness can be tuned using a thicker material for the spring layer and/or encapsulation/integration with soft materials. The empirical validation shows that the sensor has the sensitivity and the resolution necessary to discern the biologically relevant forces in a simulated cannulation task.
PDF J. B. Gafford, S. Russo, T. Ranzani, R. J. Wood, and C. J. Walsh, “
Snap-On Robotic Wrist Module for Enhanced Dexterity in Endoscopy,” in
IEEE International Conference on Robotics and Automation (ICRA), Stockholm, Sweden, 2016, pp. 4398-4405.
Publisher's VersionAbstractBurgeoning transendoscopic procedures, such as endoscopic submucosal dissection (ESD), provide a promising means of treating early-stage gastric neoplasia in a minimally-invasive way. However, the remote locations of these lesions, coupled with their origination in the submucosal layers of the gastrointestinal tract, often lead to extreme technical, cognitive and ergonomic challenges which combat the widespread applicability and adoption of these techniques. Among these challenges is achieving the in vivo dexterity required to retract and dissect tissue. By leveraging workspace and force data obtained through clinical studies, we developed a modular, disposable, distally-mounted actuator (an 'active endcap') that can augment an endoscopist's distal dexterity in ways that are not achievable with the endoscope's built-in degrees-of-freedom. The device consists of a flexible articulating 'exoskeleton' manufactured via printed-circuit MEMS (PCMEMS) which engages and deflects electrosurgical tools that are passed through the endoscopic working channel. Embedded proprioceptive sensing is implemented on-board using distributed LED/phototransistor pairs and the principle of light intensity modulation (LIM). The distal degree-of-freedom is actuated using shape memory alloy (SMA) technology, and the actuation transmission system is fully contained within a 1-inch-long end cap that can be mounted on the distal end of the endoscope, thereby obviating the need for a mechanical connection to a proximal source. Proof-of-concept tests demonstrate that the actuator adds over 50 degrees of distal articulation to existing tools and can generate 450 mN of lateral force which has been clinically determined to be sufficient for performing circumferential incisions in ESD.
PDF S. Russo, T. Ranzani, J. B. Gafford, C. J. Walsh, and R. J. Wood, “
Soft pop-up mechanisms for micro surgical tools: design and characterization of compliant millimeter-scale articulated structures,” in
IEEE International Conference on Robotics and Automation (ICRA), Stockholm, Sweden, 2016, pp. 750-757.
Publisher's VersionAbstractThis paper introduces a manufacturing technique which enables the integration of soft materials and soft fluidic micro-actuators in the Pop-up book MEMS paradigm. Such a technique represents a promising approach to the design and fabrication of low cost and scalable articulated mechanisms provided with sensing capabilities and on-board actuation with potential applications in the field of minimally invasive surgery. Design and integration of soft components in the rigid-flex laminates is described along with the resulting soft pop-up mechanisms realized at different scales. Prototype characterization is presented, demonstrating forces and dexterity in a range suitable for surgical applications, as well as the possibility to integrate sensing capabilities. Based on these results, a multi-articulated robotic arm is fabricated and mounted on top of an endoscope model to provide a proof of concept of simple robotic mechanisms that could be useful in a surgical scenario.
PDF