Controlling a bio-inspired miniature blimp using a depth sensing neural-network camera.

Miniature blimps are lighter-than-air vehicles which have become an increasingly common unmanned aerial system research platform due to their extended endurance and collision tolerant design. The UNSW-C bio-inspired miniature blimp consists of a 0.5 m spherical mylar envelope filled with helium. Four fins placed along the equator provide control over the three translatory axes and yaw rotations. A gondola attached to the bottom of the blimp contains all the electronics and flight controller. Here, we focus on using the UNSW-C blimp as a platform to achieve autonomous flight in GPS-denied environments. The majority of unmanned flying systems rely on GPS or multi-camera motion capture systems for position and orientation estimation. However, such systems are expensive, difficult to set up and not compact enough to be deployed in real environments. Instead, we seek to achieve basic flight autonomy for the blimp using a low-priced and portable solution. We make use of a low-cost embedded neural network stereoscopic camera (OAK-D-PoE) for detecting and positioning the blimp while an onboard inertia measurement unit was used for orientation estimation. Flight tests and analysis of trajectories revealed that 3D position hold as well as basic waypoint navigation could be achieved with variance (<0.1 m). This performance was comparable to that when a conventional multi-camera positioning system (VICON) was used for localizing the blimp. Our results highlight the potentially favorable tradeoffs offered by such low-cost positioning systems in extending the operational domain of unmanned flight systems when direct line of sight is available.

Development and Characterization of Novel Conductive Sensing Fibers for In Vivo Nerve Stimulation.

Advancements in electrode technologies to both stimulate and record the central nervous system's electrical activities are enabling significant improvements in both the understanding and treatment of different neurological diseases. However, the current neural recording and stimulating electrodes are metallic, requiring invasive and damaging methods to interface with neural tissue. These electrodes may also degrade, resulting in additional invasive procedures. Furthermore, metal electrodes may cause nerve damage due to their inherent rigidity. This paper demonstrates that novel electrically conductive organic fibers (ECFs) can be used for direct nerve stimulation. The ECFs were prepared using a standard polyester material as the structural base, with a carbon nanotube ink applied to the surface as the electrical conductor. We report on three experiments: the first one to characterize the conductive properties of the ECFs; the second one to investigate the fiber cytotoxic properties in vitro; and the third one to demonstrate the utility of the ECF for direct nerve stimulation in an in vivo rodent model.

Predicting Outcomes of Hormone and Chemotherapy in the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) Study by Biochemically-inspired Machine Learning.

Genomic aberrations and gene expression-defined subtypes in the large METABRIC patient cohort have been used to stratify and predict survival. The present study used normalized gene expression signatures of paclitaxel drug response to predict outcome for different survival times in METABRIC patients receiving hormone (HT) and, in some cases, chemotherapy (CT) agents. This machine learning method, which distinguishes sensitivity vs. resistance in breast cancer cell lines and validates predictions in patients; was also used to derive gene signatures of other HT  (tamoxifen) and CT agents (methotrexate, epirubicin, doxorubicin, and 5-fluorouracil) used in METABRIC. Paclitaxel gene signatures exhibited the best performance, however the other agents also predicted survival with acceptable accuracies. A support vector machine (SVM) model of paclitaxel response containing genes  ABCB1, ABCB11, ABCC1, ABCC10, BAD, BBC3, BCL2, BCL2L1, BMF, CYP2C8, CYP3A4, MAP2, MAP4, MAPT, NR1I2, SLCO1B3, TUBB1, TUBB4A, and TUBB4B was 78.6% accurate in predicting survival of 84 patients treated with both HT and CT (median survival ≥ 4.4 yr). Accuracy was lower (73.4%) in 304 untreated patients. The performance of other machine learning approaches was also evaluated at different survival thresholds. Minimum redundancy maximum relevance feature selection of a paclitaxel-based SVM classifier based on expression of genes  BCL2L1, BBC3, FGF2, FN1, and  TWIST1 was 81.1% accurate in 53 CT patients. In addition, a random forest (RF) classifier using a gene signature ( ABCB1, ABCB11, ABCC1, ABCC10, BAD, BBC3, BCL2, BCL2L1, BMF, CYP2C8, CYP3A4, MAP2, MAP4, MAPT, NR1I2,SLCO1B3, TUBB1, TUBB4A, and TUBB4B) predicted >3-year survival with 85.5% accuracy in 420 HT patients. A similar RF gene signature showed 82.7% accuracy in 504 patients treated with CT and/or HT. These results suggest that tumor gene expression signatures refined by machine learning techniques can be useful for predicting survival after drug therapies.