8 research outputs found
Aggressive Aerial Grasping using a Soft Drone with Onboard Perception
No full textContrary to the stunning feats observed in birds of prey, aerial manipulation and grasping with flying robots still lack versatility and agility. Conventional approaches using rigid manipulators require precise positioning and are subject to large reaction forces at grasp, which limit performance at high speeds. The few reported examples of aggressive aerial grasping rely on motion capture systems, or fail to generalize across environments and grasp targets. We describe the first example of a soft aerial manipulator equipped with a fully onboard perception pipeline, capable of robustly localizing and grasping visually and morphologically varied objects. The proposed system features a novel passively closing tendon-actuated soft gripper that enables fast closure at grasp, while compensating for position errors, complying to the target-object morphology, and dampening reaction forces. The system includes an onboard perception pipeline that combines a neural-network-based semantic keypoint detector with a state-of-the-art robust 3D object pose estimator, whose estimate is further refined using a fixed-lag smoother. The resulting pose estimate is passed to a minimumsnap trajectory planner, tracked by an adaptive controller that fully compensates for the added mass of the grasped object. Finally, a finite-element-based controller determines optimal gripper configurations for grasping. Rigorous experiments confirm that our approach enables dynamic, aggressive, and versatile grasping. We demonstrate fully onboard vision-based grasps of a variety of objects, in both indoor and outdoor environments, and up to speeds of 2.0 m/s— the fastest vision-based grasp reported in the literature. Finally, we take a major step in expanding the utility of our platform beyond stationary targets, by demonstrating motion-capture-based grasps of targets moving up to 0.3 m/s, with relative speeds up to 1.5 m/s.
Video Attachment: https://www.youtube.com/watch?v=HF4M7TooqfES.M
Sampling tool concepts for Enceladus lander in-situ analysis
No full textA potential future in-situ lander mission to the surface of Enceladus could be the lowest cost mission to determine if life exists beyond Earth since material from the subsurface ocean, where the presence of hydrothermal activity has been strongly suggested by the Cassini mission, is available on its surface after being ejected by plumes and then settling on the surface. In addition, the low radiation environment of Enceladus would not significantly alter the chemical makeup of samples recently deposited on the surface. A study was conducted to explore various sampling devices that could be used by an in-situ lander mission to provide 1cc to 5cc volume samples to instruments. In addition to temperature and vacuum environmental conditions, the low surface gravity of Enceladus (1% of Earth gravity) represents a new challenge for surface sampling that is not met by sampling systems developed for microgravity (e.g., comets and asteroids) or higher gravity (e.g., Europa 13%g, Moon 16%g, or Mars 38%g) environments. It is desired to acquire surface plume material that has accumulated in the top 1cm to ensure acquisition of the least processed material. Several sampling devices were developed or adapted and then tested in simulated conditions that resemble the Enceladus surface properties. These devices and test results are presented in this paper
Grasping Static and Moving Targets with a Soft Drone: Control and Prediction
No full textThis thesis studies the problem of grasping static and moving targets with a Soft Drone, which is a quadrotor platform where the landing gear is replaced with a soft manipulator. Whereas traditional rigid aerial manipulators are intolerant to positioning errors and susceptible to large contact forces, the Soft Drone leverages a soft, tendon-actuated gripper, whose inherent compliance provides robustness against imprecision and mitigates disturbances. However, the Soft Drone still requires the design of control algorithms for grasping as well as prediction methods to track a moving target.
The first part of the thesis focuses on control algorithms for grasping a static target with specific consideration towards real-world conditions. Unmodeled external disturbances, such as the added mass of the target post-grasp, are estimated and compensated through an adaptive controller. When a motion capture system is not available, the target is localized using purely onboard sensors through a perception-based approach. Experimental results are presented which show that the adaptive control scheme is capable of asymptotically stabilizing the zero tracking error of a static grasp trajectory, despite external disturbances. Initial results of the perception-based grasp are also shown using a photo-realistic simulator.
The second part of the thesis focuses on grasping a moving target. Estimation and prediction of the target’s state become key considerations when the target is moving. We employ an Extended Kalman filter for state estimation and use regularized polynomial fitting to predict the target’s future trajectory. This information is used in a linear model predictive controller, which enables the quadrotor to track the target’s state with minimal error. Simulation results show that our estimation and prediction approaches are robust against varying levels of noise and target predictability and that the control design enables successful grasping of a moving target.M.Eng
The Dual-Rasp Sampling System for an Enceladus Lander
No full textThe Dual-Rasp sampling system has been developed for the unique sampling environment of a lander mission to the surface of Saturn's moon Enceladus. Plume material from the subsurface ocean that has fallen to the surface is desired resulting in an objective to sample the topmost layer of icy material. The low gravity and potential large range of surface properties are challenges for the sampling system. The Dual-Rasp sampling system has two counter-rotating rasp cutters with teeth that remove material that is thrown up between the cutters. Two prototypes of the Dual-Rasp sampling system were built and tested, one with a carousel and one that uses pneumatics for sample transfer
Recommended from our members
Optimizing the Therapeutic Index for Medulloblastoma by Targeting Apoptosis
No full textMedulloblastoma (MB) is the most common central nervous system (CNS) malignancy affecting children, accounting for 20% of all childhood brain tumors. While current treatments, which include surgery, chemotherapy, and external beam radiation therapy, are effective at eliminating tumors in the majority of patients, there are long term consequences of these treatments. Namely, the use of ionizing radiation in the CNS of pediatric patients has been shown to lead to loss of cognitive function and neurodevelopmental delay, in addition to neurosensory impairment and pituitary-hypothalamic dysfunction. While there are multiple proposed mechanisms behind this radiation induced brain injury, the apoptotic death of healthy brains cells has been shown to be a large contributing factor. Previous studies have demonstrated that genetic knock out of BAX, a protein necessary for initiating mitochondrial apoptosis, protects neural cells from radiation induced apoptosis. However, clinically relevant pharmacological inhibitors of BAX have yet to be developed and are unlikely to be useful in this setting due to potential reductions in cure rates. We therefore sought alternative approaches to reducing BAX levels in healthy brain tissue. BAX expression in the immature brain is mediated by the growth-associated transcription factor MYC, which is also dysregulated in a majority of cancers and is a demonstrated driver of pediatric MB. We therefore hypothesized that targeting the transcription of BAX in young brain tissues by inhibiting the transcription factor MYC may protect healthy neural cells from ionizing radiation, while still eliminating MB cells.
As such, BET inhibitors JQ1 and BMS-986158 were used to treat human MB cell lines, primary murine neural cells, and C57BL/6 mice (healthy and tumor bearing). The resulting effects in combination with ionizing radiation (from a Cs-137 source or from a constant voltage X-ray source) were assessed. We found that BET inhibition led to a decrease in BCL-XL in Group 3 MB cells, and a subsequent increase in dependence on MCL-1. On the other hand, we found that BET inhibition decreases apoptotic sensitivity of healthy neural cells by downregulating BAX and BAK. These divergent effects broadened the therapeutic window for radiation treatment, leading to increased loss of MB viability while protecting healthy neural cells from radiation induced apoptosis. Together, these data provide insight on the differing mechanistic effects of BET inhibition on cancerous MB cells and healthy brain cells, and the enhancement of the therapeutic window for treatment of pediatric patients diagnosed with medulloblastoma
A Certifiable Algorithm for Simultaneous Shape Estimation and Object Tracking
No full textApplications from manipulation to autonomous vehicles rely on robust and general object tracking to safely perform tasks in dynamic environments. We propose the first certifiably optimal category-level approach for simultaneous shape estimation and pose tracking of an object of known category (e.g. a car). Our approach uses 3D semantic keypoint measurements extracted from an RGB-D image sequence, and phrases the estimation as a fixed-lag smoothing problem. Temporal constraints enforce the object\u27s rigidity (fixed shape) and smooth motion according to a constant-twist motion model. The solutions to this problem are the estimates of the object\u27s state (poses, velocities) and shape (paramaterized according to the active shape model) over the smoothing horizon. Our key contribution is to show that despite the non-convexity of the fixed-lag smoothing problem, we can solve it to certifiable optimality using a small-size semidefinite relaxation. We also present a fast outlier rejection scheme that filters out incorrect keypoint detections with shape and time compatibility tests, and wrap our certifiable solver in a graduated non-convexity scheme. We evaluate the proposed approach on synthetic and real data, showcasing its performance in a table-top manipulation scenario and a drone-based vehicle tracking application.Accepted to IEEE RA-L. 11 pages, 6 figures (with appendix). Code released at https://github.com/MIT-SPARK/certifiable_tracking. Video available at https://youtu.be/eTIlVD9pDt
Aggressive Aerial Grasping using a Soft Drone with Onboard Perception
No full textContrary to the stunning feats observed in birds of prey, aerial manipulation
and grasping with flying robots still lack versatility and agility.
Conventional approaches using rigid manipulators require precise positioning
and are subject to large reaction forces at grasp, which limit performance at
high speeds. The few reported examples of aggressive aerial grasping rely on
motion capture systems, or fail to generalize across environments and grasp
targets. We describe the first example of a soft aerial manipulator equipped
with a fully onboard perception pipeline, capable of robustly localizing and
grasping visually and morphologically varied objects. The proposed system
features a novel passively closing tendon-actuated soft gripper that enables
fast closure at grasp, while compensating for position errors, complying to the
target-object morphology, and dampening reaction forces. The system includes an
onboard perception pipeline that combines a neural-network-based semantic
keypoint detector with a state-of-the-art robust 3D object pose estimator,
whose estimate is further refined using a fixed-lag smoother. The resulting
pose estimate is passed to a minimum-snap trajectory planner, tracked by an
adaptive controller that fully compensates for the added mass of the grasped
object. Finally, a finite-element-based controller determines optimal gripper
configurations for grasping. Rigorous experiments confirm that our approach
enables dynamic, aggressive, and versatile grasping. We demonstrate fully
onboard vision-based grasps of a variety of objects, in both indoor and outdoor
environments, and up to speeds of 2.0 m/s -- the fastest vision-based grasp
reported in the literature. Finally, we take a major step in expanding the
utility of our platform beyond stationary targets, by demonstrating
motion-capture-based grasps of targets moving up to 0.3 m/s, with relative
speeds up to 1.5 m/s
Restoration of tumour-growth suppression in vivo via systemic nanoparticle-mediated delivery of PTEN mRNA
No full text© 2018, The Author(s). Phosphatase and tensin homologue deleted on chromosome 10 (PTEN) is a well-characterized tumour-suppressor gene that is lost or mutated in about half of metastatic castration-resistant prostate cancers and in many other human cancers. The restoration of functional PTEN as a treatment for prostate cancer has, however, proven difficult. Here, we show that PTEN messenger RNA (mRNA) can be reintroduced into PTEN-null prostate cancer cells in vitro and in vivo via its encapsulation in polymer–lipid hybrid nanoparticles coated with a polyethylene glycol shell. The nanoparticles are stable in serum, elicit low toxicity and enable high PTEN mRNA transfection in prostate cancer cells. Moreover, significant inhibition of tumour growth is achieved when delivered systemically in multiple mouse models of prostate cancer. We also show that the restoration of PTEN function in PTEN-null prostate cancer cells inhibits the phosphatidylinositol 3-kinase (PI3K)–AKT pathway and enhances apoptosis. Our findings provide proof-of-principle evidence of the restoration of mRNA-based tumour suppression in vivo
