This study proposes a novel hybrid underwater robot platform, HERO-BLUE, that integrates swimming and legged motions. HERO-BLUE is equipped with a multimodal fin comprising numerous passive joints that can act as both a swimming fin and a walking limb. This multimodal fin is integrated with a salamander-like spine and soft undulating fin in a single robot system to enable three representative locomotion modes: swimming, walking, and crawling. This paper provides details of the hardware configuration of HERO-BLUE, mathematical models for each locomotion mode, motion planning methodology, and control strategies for enabling hybrid motions. For legged motions, a dynamic simulation study was conducted to analyze the ground reaction force. The results verified the effectiveness of the proposed multimodal fin. In water tank experiments, the swimming capability was validated through thrust tests and swimming trials, and the legged motion capability was quantitatively validated in three environments, including gravel, water flow, and slopes. Furthermore, field trials were conducted in real-life scenarios in open seas, lakes, and stream beds. The experimental results show that HERO-BLUE can successively combine swimming and legged motion capability and increase underwater mobility in various challenging underwater environments.

The supply of IoT devices is increasing year by year. Even in industries that demand sophistication, such as unmanned driving, construction, and robotics industry, IoT devices are being utilized. However, the security of IoT devices is lagging behind this development due to their diverse types and challenging firmware execution environments. The existing methods, such as direct device connectivity or partial emulation, are used to solve this. However, full system emulation is better suited for the large-scale analysis, because it can test many firmwares without requiring devices. Therefore, recent studies have integrated emulation and software testing techniques such as fuzzing, but they are still unsuitable for testing various firmware and inefficient. In this poster, we propose FirmColic, which combines fuzzing with concolic execution to mitigate these limitations. FirmColic is a type of augmented process emulation, which improves the effectiveness of fuzzing using keyword extraction based on concolic execution. Also, we apply five arbitration techniques in an augmented process emulation environment for the high success rates of the emulation. We prove that FirmColic has faster detection, more crash detection, and a higher code coverage than the previous studies.

Searching top-k nearest neighbor (kNN) based on the vector similarity is a common problem in many domains. Unlike approximate kNN search, exact kNN needs to check a large portion of the dataset, if not the entire set. When the data volume is large, the dataset cannot fit in the main memory and has to be stored on a storage device such as flash drive. ES4D is a kNN search platform implemented near-data on the solid-state drive (SSD). ES4D accelerates kNN search by using two levels of early termination, using pre-clustering of the dataset vectors and vector sharding. ES4D incorporates several optimization techniques to aid such early terminations. Using near-data processing, ES4D further enhances performance and reduces energy consumption. Performing the kNN search on the SSD side not only improves energy efficiency but also enables ES4D to optimize the physical page placement of dataset vectors on SSD to maximize the search throughput. We evaluated ES4D using various vector datasets. ES4D achieves 1.9× search performance improvement over existing exact kNN search mechanisms. Also, by off-loading the distance calculation to the SSD side, ES4D reduces energy consumption for kNN search by 8.1×.

Hybrid fuzzers combine both fuzzing and concolic execution with the wish that the fuzzer will quickly explore input spaces and the concolic execution will solve the complex path conditions. However, existing hybrid fuzzers such as Driller cannot be effectively directed, for instance, towards unsafe system calls or suspicious locations, or towards functions in the call stack of a reported vulnerability that we wish to reproduce. In this poster, we propose DrillerGO, a directed hybrid fuzzing system, to mitigate this problem. It mainly consists of a static analysis and a dynamic analysis module. In the static analysis, it searches suspicious API call strings in the recovered control flow graph (CFG). After targeting some suspicious API call lines, it runs the concolic execution along with path guiding. The path guiding is helped by backward pathfinding, which is a novel technique to find paths backward from the target to the start of main(). Also, we will show that DrillerGo can find the crashes faster than Driller through experimental results.

We designed a manipulation purpose small agent vehicle and performed a ghost net recovery mission with the vehicle to test its control accuracy and capacity to be used in other tasks. We constructed a control model that allows the vehicle to perform the complex motion of 4-degrees-of-freedom (DOF) movements and underwater manipulation. Also, we implemented a ghost net recovery algorithm that can automatically detect, grip, and lift ghost nets. The test results show stable yaw, depth, and position control capabilities, and the manipulation performance was verified by automatically performing a ghost net recovery mission.

Sonar sensors are widely used for underwater observations because they can be used in a turbid stream and have a long operating range. However, images taken with sonar sensors are difficult to identify because of low resolution. In this paper, we proposed a method based on a generative adversarial network to increase the resolution of the underwater sonar image. We built the network of 16 residual blocks and eight convolutional layers. We then trained it with sonar images cropped in several ways. As a result, we could improve the resolution of sonar images from various scenes and recorded a higher peak signal-to-noise ratio than interpolation. The proposed method could help to identify the underwater object without losing the working range of sonar.