In building artificial intelligence (AI) agents, referring to how brains function in real environments can accelerate development by reducing the design space. In this study, we propose a probabilistic generative model (PGM) for navigation in uncertain environments by integrating the neuroscientific knowledge of hippocampal formation (HF) and the engineering knowledge in robotics and AI, namely, simultaneous localization and mapping (SLAM). We follow the approach of brain reference architecture (BRA) (Yamakawa, 2021) to compose the PGM and outline how to verify the model. To this end, we survey and discuss the relationship between the HF findings and SLAM models. The proposed hippocampal formation-inspired probabilistic generative model (HF-PGM) is designed to be highly consistent with the anatomical structure and functions of the HF. By referencing the brain, we elaborate on the importance of integration of egocentric/allocentric information from the entorhinal cortex to the hippocampus and the use of discrete-event queues.

Building a human-like integrative artificial cognitive system, that is, an artificial general intelligence (AGI), is the holy grail of the artificial intelligence (AI) field. Furthermore, a computational model that enables an artificial system to achieve cognitive development will be an excellent reference for brain and cognitive science. This paper describes an approach to develop a cognitive architecture by integrating elemental cognitive modules to enable the training of the modules as a whole. This approach is based on two ideas: (1) brain-inspired AI, learning human brain architecture to build human-level intelligence, and (2) a probabilistic generative model (PGM)-based cognitive architecture to develop a cognitive system for developmental robots by integrating PGMs. The proposed development framework is called a whole brain PGM (WB-PGM), which differs fundamentally from existing cognitive architectures in that it can learn continuously through a system based on sensory-motor information. In this paper, we describe the rationale for WB-PGM, the current status of PGM-based elemental cognitive modules, their relationship with the human brain, the approach to the integration of the cognitive modules, and future challenges. Our findings can serve as a reference for brain studies. As PGMs describe explicit informational relationships between variables, WB-PGM provides interpretable guidance from computational sciences to brain science. By providing such information, researchers in neuroscience can provide feedback to researchers in AI and robotics on what the current models lack with reference to the brain. Further, it can facilitate collaboration among researchers in neuro-cognitive sciences as well as AI and robotics.

The vastness of the design space that is created by the combination of numerous computational mechanisms, including machine learning, is an obstacle to creating artificial general intelligence (AGI). Brain-inspired AGI development; that is, the reduction of the design space to resemble a biological brain more closely, is a promising approach for solving this problem. However, it is difficult for an individual to design a software program that corresponds to the entire brain as the neuroscientific data that are required to understand the architecture of the brain are extensive and complicated. The whole-brain architecture approach divides the brain-inspired AGI development process into the task of designing the brain reference architecture (BRA), which provides the flow of information and a diagram of the corresponding components, and the task of developing each component using the BRA. This is known as BRA-driven development. Another difficulty lies in the extraction of the operating principles that are necessary for reproducing the cognitive-behavioral function of the brain from neuroscience data. Therefore, this study proposes structure-constrained interface decomposition (SCID), which is a hypothesis-building method for creating a hypothetical component diagram that is consistent with neuroscientific findings. The application of this approach has been initiated for constructing various regions of the brain. In the future, we will examine methods for evaluating the biological plausibility of brain-inspired software. This evaluation will also be used to prioritize different computational mechanisms, which should be integrated and associated with the same regions of the brain.

We propose the factorizing variational autoencoder (FAVAE), a generative model for learning dis- entangled representations from sequential data via the information bottleneck principle without supervision. Real-world data are often generated by a few explanatory factors of variation, and disentangled representation learning obtains these factors from the data. We focus on the disen- tangled representation of sequential data which can be useful in a wide range of applications, such as video, speech, and stock markets. Factors in sequential data are categorized into dynamic and static ones: dynamic factors are time dependent, and static factors are time independent. Previous models disentangle between static and dynamic factors and between dynamic factors with different time dependencies by explicitly modeling the priors of latent variables. However, these models cannot disentangle representations between dynamic factors with the same time dependency, such as disentangling “picking up” and “throwing” in robotic tasks. On the other hand, FAVAE can disentangle multiple dynamic factors via the information bottleneck principle where it does not require modeling priors. We conducted experiments to show that FAVAE can extract disentangled dynamic factors on synthetic, video, and speech datasets.

One problem in real-world applications of reinforcement learning is the high dimensionality of the action search spaces, which comes from the combination of actions over time. To reduce the dimensionality of action sequence search spaces, macro actions have been studied, which are sequences of primitive actions to solve tasks. However, previous studies relied on humans to define macro actions or assumed macro actions to be repetitions of the same primitive actions. We propose encoded action sequence reinforcement learning (EASRL), a reinforcement learning method that learns flexible sequences of actions in a latent space for a high-dimensional action sequence search space. With EASRL, encoder and decoder networks are trained with demonstration data by using variational autoencoders for mapping macro actions into the latent space. Then, we learn a policy network in the latent space, which is a distribution over encoded macro actions given a state. By learning in the latent space, we can reduce the dimensionality of the action sequence search space and handle various patterns of action sequences. We experimentally demonstrate that the proposed method outperforms other reinforcement learning methods on tasks that require an extensive amount of search.

Most of agents that learn policy for tasks with reinforcement learning (RL) lack the ability to communicate with people, which makes human-agent collaboration challenging. We believe that, in order for RL agents to comprehend utterances from human colleagues, RL agents must infer the mental states that people attribute to them because people sometimes infer an interlocutor's mental states and communicate on the basis of this mental inference. This paper proposes PublicSelf model, which is a model of a person who infers how the person's own behavior appears to their colleagues. We implemented the PublicSelf model for an RL agent in a simulated environment and examined the inference of the model by comparing it with people's judgment. The results showed that the agent's intention that people attributed to the agent's movement was correctly inferred by the model in scenes where people could find certain intentionality from the agent's behavior.

Equivalence structure (ES) extraction can allow for finding correspondence relations between different sequential datasets. A K -dimensional ES is a set of K -tuples to specify K -dimensional sequences that are considered equivalent. Whether or not two K -dimensional sequences are equivalent is decided based on comparisons of all of their subsequences. ES extraction can be used for preprocessing for transfer learning or imitation learning, as well as an analysis of multidimensional sequences. A recently proposed method called incremental search (IS) was much faster than brute-force search. However, IS can still take a long time to obtain ESs, because ESs obtained by IS can be subsets of other ESs and such subsets must be removed in the process. In this paper, we propose a new fast method called pairwise incremental search (PIS). In the process of PIS, the aforementioned problem about subsets of ESs does not exist, because the elements of ESs are searched pairwise. As shown by results of two experiments we conducted, PIS was 48 times faster than IS in an experiment using synthetic datasets and 171 times faster in an experiment using motion capture datasets.

In cooperation, the workers must know how co-workers behave. However, an agent's policy, which is embedded in a statistical machine learning model, is hard to understand, and requires much time and knowledge to comprehend. Therefore, it is difficult for people to predict the behavior of machine learning robots, which makes Human Robot Cooperation challenging. In this paper, we propose Instruction-based Behavior Explanation (IBE), a method to explain an autonomous agent's future behavior. In IBE, an agent can autonomously acquire the expressions to explain its own behavior by reusing the instructions given by a human expert to accelerate the learning of the agent's policy. IBE also enables a developmental agent, whose policy may change during the cooperation, to explain its own behavior with sufficient time granularity.

An equivalence structure (ES) is a set of K-tuples composed of IDs that indicate K D sequences considered similar. The similarity is calculated by comparing continuous subsequences taken from KD sequences. When comparing one of such continuous subsequences with that of a different tuple, the starting point of the former subsequence can differ from the latter subsequence. The extraction of ES's can be applied for various purposes, such as feature extraction, the analysis of neural network behavior, and imitation learning. However, the extraction requires comparisons of all possible KD sequences taken from N sequences; that is, the number of the possible sequences is the number of K-permutations of N. Therefore, the comparisons cause combinatorial explosion with K. In this paper, we propose a new method to extract ES's increasing K, and show that, in our experiments, our method allowed us to extract ES's much faster than the brute-force search, which is the only existing method.

We developed a four-arm four-crawler advanced disaster response robot called OCTOPUS. Disaster response robots are expected to be capable of both mobility, e.g., entering narrow spaces over very rough unstable ground, and workability, e.g., conducting complex debris-demolition work. However, conventional disaster response robots are specialized in either mobility or workability. Moreover, strategies to independently enhance the capability of crawlers for mobility and arms for workability will increase the robot size and weight. To balance environmental applicability with the mobility and workability, OCTOPUS is equipped with a mutual complementary strategy between its arms and crawlers. The four arms conduct complex tasks while ensuring stabilization when climbing steps. The four crawlers translate rough terrain while avoiding toppling over when conducting demolition work. OCTOPUS is hydraulic driven and teleoperated by two operators. To evaluate the performance of OCTOPUS, we conducted preliminary experiments involving climbing high steps and removing attached objects by using the four arms. The results showed that OCTOPUS completed the two tasks by adequately coordinating its four arms and four crawlers and improvement in operability needs.

It is important to understand the mechanism of human standing-up motion to improve the declined physical ability of the elderly people. This study employs the concept of muscle synergies (modular structure of coordinative muscle activation) to understand how humans coordinate their muscles to achieve the standing-up motion. Neuro-musculoskeletal model was developed to represent human body to generate standing-up motion. Using the developed model, forward dynamic simulation was used to analyze how humans utilized the muscle synergies to realize the motion. Results showed that the developed model could generate the standing-up motion with four muscle synergies rather than controlling individual muscles. Moreover, further analysis showed that three different strategies of the standing-up motion could be generated only by changing the start time of the particular muscle synergy.

The standing-up motion is one of the most important activities of daily livings. In order to understand the strategy to achieve the standing-up motion, muscle synergy analysis is applied to the measured data during human standing-up motion. In addition, musculoskeletal model which consists of three body segments and nine muscles in lower limb is developed to ensure that the standing-up motion can be generated by muscle synergies. As a result, three muscle synergies have been extracted from the human standing-up motion, and each synergy strongly corresponded to characteristic kinematic events: momentum flexion, momentum transfer, and posture stabilization. Results of forward dynamic simulation show that the standing-up motion can be achieved by controlling time-varying weighting coefficient of three muscle synergies instead of controlling individual nine muscles.

Although standing-up motion is an important activity of daily living, it remains unclear how people perform the motion in different situations. As described in this paper, muscle synergy analysis is applied to standing-up motions performed at different circumstances, such as two different heights and at three different speeds. Results elucidated three invariant groups of synchronized muscle activations: The first synergy pulls the ankle and raises the hip. The second synergy extends the upper body. The third synergy stabilizes posture. Results also show that people controlled the activation coefficient of each synergy differently during all motions. The slower the standing-up motion is, the longer each synergy activates to adapt to the slower motion speed. Results of this study show that people use the same group of synchronized muscle activation and only control the activation coefficient to achieve adaptive standing-up motion.

As the population continues to age, the number of elderly people requiring healthcare is increasing. In order to improve their physical function, they need to get physical training. There are activities which require integrated arm movements and lower body movements. However there are no quantitative testing methods of the degree of recovery for the coordination between arm movements and lower body movements. In this study, we focus on dart throwing motion as arm movements in lower body movements and suggest the quantitative evaluation of the coordination between arm movements and lower body movements in dart throwing motion. Normalized correlation coefficient (NCC) between arm and lower body was computed at different throwing distances. In addition the standard deviation of the NCC was computed in order to investigate the stability of the joint correlation evaluation. This analysis shows that the correlation between elbow and ankle, or between elbow and knee, are increased at throwing long distance. We suggest that the NCC between elbow angle and right knee angle may be used for the evaluation of the joint correlation between arm movements and lower body movements in dart throwing motion.

Recently healthcare of the elderly people has become a serious issue in medical and rehabilitation areas. In order to know their functional mobility and provide sufficient medical treatment, it is important to measure their body state precisely and objectively. Therefore we developed a wearable and wireless sensor of gyroscope and accelerometer (PocketIMU2) as an easy and precise measurement of human motions. In the sensor, we employed a small and high accurate LiNbO3 crystal to achieve joint angle computation with simple integration of angular velocity. In the current paper, we evaluate the accuracy of the sensor in two important basic motion, such as a walking and sit-to-stand motions. Computed joint angles of ankle, knee, and hip are compared to the reference data measured from a optical motion capture system in term of coefficients of correlation and root mean square error. As a result, coefficient of correlation showed very high value for all joint angles, and root mean square error was adequately small. This strongly supports the usage of our developed gyroscope and accelerometer sensor for monitoring human body movement for medical usage.

This paper proposes the planning and navigation of a surgical robot to perform safe and effective operations using finite element analysis results based on physical models of the brain. The physical brain models were proposed based on the results of physical property tests. Finite element analyses of virtual tension tests were carried out under the same conditions as the actual tension tests, and the results of the analysis coincided with those of the test. The finite element analyses of robotic brain surgery were carried out using a two-dimensional structure, which consisted of the proposed physical models. Assuming two types of robot models are used (the brain spatula and capsule type robot respectively), stress distribution in the brain, caused by each robot model, was simulated in the analyses. The preoperative planning was considered based on the analysis results, and the safe velocity of the robot’s movement was estimated from the same. The results of this research showed that organ modeling was necessary for preoperative planning in order to implement safe surgery.

This paper proposes the concept of "agent networks" to realize flexible and unified integration of multi-agent intelligent robot systems. As an example, we developed a hand-to-hand robot on the agent network. Its robust and flexible behaviour proved the effectiveness of the method.

For effective recognition, a recognition structure that controls the information flow among the specialized processing modules should reflect the implicit correlation structure of the environmental input. Autonomous construction of a recognition structure will lead to extensive improvement in the flexibility of the adaptive recognition system. For this purpose we propose a matchability-oriented feature selection that can collect highly correlated features at each local module. Conventional techniques, which collect features that are more independent, are not suitable. Matchability is a concept derived from the recognition functions of an adaptive intelligent agent (useful for action generation) and corresponds to the probability of input data items matching stored data items in the recognition system. The proposed algorithm changes the weights attached to each feature depending on the degree of matchability of each feature. This algorithm could select highly correlated features in simple simulation.