An experimental and theoretical study of real-time robot balancing on inclined surfaces with electrical feedback circuitry is presented. Force sensors are experimentally shown to extend the sustainability of a stable robot posture beyond a critical surface inclination. For this purpose, the inclination feedback from the force sensors is used to adjust the robot’s ankle-pitch-motor angle above the critical inclination, thus enabling the maintenance of a stable robot posture. Further, the Inverted Pendulum Model (IPM) 43-45 is extended to the case of inclined surfaces. Through application of this extended IPM it is demonstrated, that simultaneous use of gyro-sensor data can minimize the necessary initial adjustment of the motor angle for controlled robot-body rotation, which additionally has the positive effect of reducing possible overshoots of the motor’s rotation angle during feedback. Consequently, the reported feedback control improves the robot-body stability on inclined surfaces. Efficient implementation of the developed control scheme into an existing robot’s electrical system is proposed.

In this paper, we propose a surface-identification system for stable humanoid-robot walking on various types of surfaces using force sensors mounted under the robot feet. For experimental identification analysis of the surface condition, we measured the sensor-output data using five different types of test surfaces. To achieve fast dynamic recognition capability of changing surface conditions, we applied an overlapped sliding-window method for the incoming sensor-data stream to generate dynamically four distinguishable well-known features from the raw sensor data. The multi-class k-nearest-neighbor (MC-kNN) classifier rather than a binary classifier is used for online classification of the measured robot-walking pattern and classification-accuracy evaluation. Further, we combine the four studied feature descriptors into a fused multi-feature descriptor rather than invoking each feature descriptor independently, increasing the classification performance. Our analysis results verify that 90.4% maximum overall accuracy with 91.49% average precision can be achieved, demonstrating the realization of a better cost-performance trade-off than in other previous research works. The obtained results are useful for balancing the robot body through optimized controlling of the robot motors according to the recognized different surfaces during robot motion.

Real-time surface recognition has become a critical factor for ensuring safe walking of intelligent biped robots in a complex human living environment. This work aims at enabling wide cost-efficient implementation of sensing solutions for surface recognition via walking-pattern classification by restricting the necessary hardware to a cost-economic microprocessor and a single type of force sensors. For experimental analysis, we explored the walking-pattern classification performance using a framework which combines a support vector machine (SVM) and four time-domain feature descriptors, i.e., mean of amplitude (MA), integral of absolute value (IAV), variance (VAR), and root mean square (RMS). During the online pattern classification, the dynamical force-sensory-data stream was extracted using a real-time overlapped-window-based method. Multiple binary SVM classifiers were applied for solving the multi-class classification problem, due to the reasonably high accuracy and the relatively small complexity for hardware implementation, allowing simultaneous strength exploitation of above four individual feature descriptors with a one-versus-one (OVO) strategy. The experimental results, obtained with 250 samples/surface, verified 93.8% mean average precision, 93.7% average accuracy and recall rates of 98.8%, 91.6%, 82.0%, 98.0%, 98.0% for smooth wood, rough foam, smooth foam, thick carpet, and thin carpet, respectively. Only the dynamical force-sensing data were employed for a 10-fold cross validation, which enabled the high processing speed of 0.73 ms/stride. The developed cost-efficient and accurate surface-recognition system can be useful for ensuring safe in-door locomotion for the biped robot and can help the robot to better understand the human living environment by increasing its sensing diversity.

We report the development of a biped-robot system with real-time surface recognition and walking-speed adjustment to control the robot motion during walking on different types of surfaces. Four types of test surfaces (i.e. rough foam (RF), smooth foam (SF), thin carpet (TC) and smooth table (ST)) are considered in the system verification. For surface-property recognition we use ultra-thin-membrane force sensors, mounted under the robot feet, and a classification circuit, implemented on an Arduino Uno board. The walking-speed adjustment is performed with an external control circuit, which receives the surface-recognition signal from the classification circuit and sends a feedback signal to the robot controller (i.e. RCB-4HV) for adjusting the walking speed accordingly. We applied the nearest-neighbor-classification algorithm with the Euclidean-distance measure and a set of reference data, to distinguish between the four test surfaces based on the robot's real-time walking pattern. The mean absolute value (MAV) feature descriptor is used to generate four different types of reference walking pattern, corresponding to the four different surfaces. In our experiments it is observed, that the ST surface performs best in terms of average surface-recognition latency (SRL) (~3.6 sec) during walking on same surface. On the other hand, the surface transition from TC to SF showed minimum surface-transition latency (STL) (~8.2 sec) with correct speed change from 135 to 160 robot-motor-configuration frames per stride (frames/stride), while the transition from SF to TC surfaces showed maximum STL (~11.6 sec) including speed change from 160 to 135 frames/stride. The obtained results are useful for development of the next generation of surface-recognition and speed-adjustment systems, implemented in humanoid robots to enable balanced and stable walking in environments with multiple changed surface properties.

We report an energy-efficiency analysis for the walking pattern of a humanoid robot on different indoor surfaces with different walking speeds. The walking efficiency is measured through experiments for the maximum distance, which can be covered by the robot following specific walking patterns. For this purpose, we developed an energy-measurement-circuit (EMC) to measure power and energy consumption. Two different walking surfaces (i.e. hard and soft-surfaces) and three different walking speeds (i.e. 220 frames/stride (slow-speed), 190 frames/stride (medium-speed) and 160 frames/stride (fast-speed)) were used. The walking pattern was generated by the robot-operating-software platform (ROSP) and the robot controller (i.e. RCB-4HV). Pizo-resistive-membrane force sensors (PRMFS) below the robot feet were used for walking-pattern recording. From the measurement data, it is observed that the humanoid robot with one battery charge can cover on the hard-surface maximum distances of 67.3 m for slow-speed, 77.07 m for medium-speed and 96.24 m for fast-speed. In comparison, the maximum distances on the soft-surface are only 36.94 m for slow-speed, 44.07 m for medium-speed and 55.23 m for high-speed, meaning about 80% higher energy consumption for a given identical distance. It is also observed, that the energy consumption during walking on the hard-surface for 1-meter distance covered (i.e. 181.19 J for slow-speed, 171.13 J for medium-speed and 166.68 J for fast-speed) is comparatively lesser than on the soft-surface (i.e. 365.78 J for slow-speed, 325.23 J for medium-speed and 310.15 J for fast-speed). Our experiments show, that the energy consumption (in %) during walking is substantially smaller on hard surfaces than on soft surfaces, namely, 50.46% for slow-speed, 47.38% for medium-speed and 46.25% for fast-speed. It is further shown, that the fast-speed-walking pattern on a hard surface has the highest energy efficiency among the six analyzed walking conditions. The obtained results are useful for energy-efficient walking-pattern recognition in future-generation artificial-intelligence-enabled humanoid-robot design.

This work presents our development of autonomous walking robot control using machine learning algorithm. We have investigated sensor driven walking robot movement to develop supervised learning based control algorithm using neural network methods. We used robots hardware data such as pressure sensor data for accurate neural network (NN) classification. The analyzed result shows that ~25-30 numbers of hidden neurons will perform the best result in terms of mean square error (mse), error-gradient, learning time and regression for small scale data analysis. The analyzed results are useful for next generation FPGA based artificial intelligence (AI) chip development for robot movement control.