Walking on flat terrains
9

One of the most significant advantages of humanoid robots with respect to other mobile robots is their ability to walk on terrains with various characteristics. Making steps without changing the upper body posture, humanoid robots are capable of passing over different ground surfaces, i.e. compliant surfaces, slippery surfaces, rough terrains, and etc. Despite the fact that walking is a very simple task for human beings, realizing walking patterns on the humanoid robots in real environment is a demanding task. This is due to the unactuated DOF between the robot feet and the ground surface which makes the robot vulnerable to lose its balance. Walking would be even a more severe task, when the humanoid robot has the height and weight similar to a mature man. For a humanoid with such specifications, so many problems arise during walking, while these problems would not exist during working with a kid-size humanoid.

SURENA III is capable of walking with various step lengths and walking speeds on different surfaces in terms of terrain characterization. The step length may change from 0 cm (on spot) to 50 cm during walking, while the maximum walking speed is 0.7 km/h. This generation of SURENA is 7 times faster than the last version, SURENA II. This walking speed is safe for the final version of the robot, while the walking speed 1 km/h has been reached by the bipedal robot without the arms and the body cover. The robot can change its speed as well as step length during walking, based on the commands received from the high level control unit.

Object manipulation

The procedure of manipulating an object may be divided into two parts. The first part which is one of the most demanding open challenges in the field of robotics deals with gripping and grasping. In this phase, after detection of the exact position of the object, a suitable strategy for griping the object should be considered. This strategy highly depends on the object’s shape, compliance, mass and inertia properties.  Although for a human being, it is a very simple task to pick various objects with different shapes and other attributes, for a robot it is a very complicated task. The second part copes with manipulation of the object and taking into account of the effects of object dynamics on the robot’s motion. In the case of manipulating heavy objects, dynamics of the object may threaten tip-over stability of mobile robots.

SURENA III’s arm has 7 DOF consisting of 3 DOF in shoulder, 1 DOF in elbow, and 3 DOFs in wrist. This redundant structure, which has been inspired from human’s arm, enables the robot to perform dexterous manipulations. Furthermore, the robot’s hand is comprised of 5 fingers, while all of them are actuated by one motor and a string-based transmission system. By such a mechanism, the robot is capable of just gripping some symmetric objects. The procedure is such that after the detection of the object and determining the object position by Kinekt, the robot estimates how many steps should be taken in order for the object to be in the workspace of the arm. Then, employing an inverse kinematics based planning approach, the optimal path for griping the object is generated.

14
Online adaptation
10

The most significant prominence of legged robots over wheeled robots is their ability to negotiate uneven terrains. This ability would be more highlighted when one considers that the environment around us has been constructed for human beings. Hence, exploiting this advantage to make the humanoid robots able to walk on various uneven terrains would be a step toward realizing the dream of employing the humanoid robots as assistants and servants in our daily life. Regarding this matter, online adaptations of humanoid robots for walking on unknown uneven surfaces is an open problem in this field.

One of the most advanced features of SURENA III humanoid robot is walking on uneven terrains with height uncertainties. In fact, the robot has the ability to negotiate flat obstacles with various unknown heights during walking. The procedure is such that soon or late landing of the robot swing foot is detected by the contact sensors that are located on the robot sole, and consistent modifications are applied to the preplanned trajectories to adapt the robot foot to the new situations. The heights of the obstacles are restricted to be not more than the maximum swing foot height during walking.

Turning on spot or a circle

SURENA III is able to walk on circles with various radiuses, from turning on spot which the radius equals zero, to walking on a direct line while the radius is infinite. The robot is capable to walk on all of the radiuses in this spectrum. In the turning on spot maneuver, the factors that restrict the arc that the robot sweeps in each step are the self-collision avoidance of the legs, as well as the joints movement limitations. In turning on a circle, for a constant walking speed, the radius of the circle, the arc in each step, and the step length are dependent. Hence, just two of them may be specified by the high level control unit and the other one is computed based on the determined walking speed.

12
Face detection
9-3

One of the modules that boosts the autonomy of the SURENA III humanoid robot is vision which is provided by Kinect. This module enables the robot to detect various objects as well as the human faces that are in its range. So, based on the data that have been obtained from vision and image processing, the robot is able to perform its tasks consistent with its surrounding environment.

walking up and down stairs

Walking on uneven terrains is one of the challenging subjects of research in the humanoid robotics field. This is not only due to the fact that the stability may be threatened during walking on such surfaces, but also because of the high amount of load that the actuators should withstand, especially for walking up the stairs. SURENA III can both ascend and descend the stairs with various depths (30 cm to 50 cm) and heights (0 cm to 10 cm). The robot can align itself in front of the staircases, based on the feedbacks obtained from Kinect and image processing.

7
walking up and down sloped terrains
15

During walking on inclined surfaces, the direction of gravity is not perpendicular to the direction of motion. Since gravity is the factor that limits the motion feasibility of the mobile robots, during walking on sloped surfaces the projection of the support polygon on the plane perpendicular to the motion direction would reduce. Hence, the more the inclination of the surface is, the harder the motion feasibility constraints satisfaction would be. SURENA III is capable of walking on sloped terrains from the inclination range of -10 to 10 degree.

Aerobatic motions

Aerobatic motions are those which are not for performing a task, but just to show some capacities of the robot. For example, when the robot squats, it proves its capability to withstand high amount of load in its joints. Furthermore, this action demonstrates high range of motion in the joints, especially knee joint. Also, some other motions have been designed to exhibit the synchronized legs and arms motion, as well as the ability of the robot to preserve its stability on one leg, under some demanding conditions.

8-2
Motion imitation and Speech recognition
11

SURENA III has the ability to generate whole body motions, and imitating the motion of a user standing in front of it. In this loop, the robot detects motion of the user, using the visual data and processing the skeleton of the user body. Then, consistent with the user action, the robot generates real time whole body actions.

SURENA III is capable of interact with the user through speech. It can detect various words in Persian and perform some predefined tasks based on the received signals. This module provides a high-level communication bridge between the robot and the user.

Shooting a ball

One of the maneuvers that demonstrates the online interaction of the SURENA III humanoid robot with its environment is shooting the ball with various sizes and compliances. In order to shoot the ball, the robot puts its foot on the ball and then brings it back to a suitable position and then shoots it. In this scenario, the robot is able to detect size of the ball based on the vertical interaction feedback between the foot and the ball. During shooting the ball, due to the reaction impact that exerts from the ball to the robot foot, the stability may be violated. So, suitable strategy for tacking with such disturbance has been considered.

13


'designed by jadooweb