AI News, Video Friday: Walking on Ceilings, Cat-Inspired Legs, and Robot Grasps Tofu

Video Friday: Walking on Ceilings, Cat-Inspired Legs, and Robot Grasps Tofu

The 2015IEEE International Conference on Intelligent Robots and Systems (IROS) ends todayin Hamburg, Germany, and we’ve heard that some 2,500 people attended the talks and visited the exhibit hall.

Also the muscle-tendon complex, composed of the gastrocnemius muscle and the Achilles’ tendon that cross the knee joint and the ankle joint contributes much to the movements such as running and jumping particularly.

For the knee and ankle joints, a four-bar linkage mechanism with one elastic linkage, in which the knee joint is driven by an electric rotary motor and the ankle joint is passive, is applied.

This paper proposes a new wheel mechanism, the swing-grouser wheel, which can climb high steps (especially in low friction environments) and has high energy efficiency.

Furthermore, the performance of the swing- grouser wheel was confirmed using a real device experiment and a 2D physics simulation, and improved using a full search of the parameters of the swing-grouser wheel.

Through its two modes of operation—Driving Mode and Tumbling Mode— this mechanism is able to both drive quickly on smooth surfaces at roughly 1.7 times desired speed and climb objects as tall as 67% of the diameter of the mechanism.

The unique gearing configuration that facilitates these dual capabilities is described, and testing quantifies that this nonprecision gearing system is roughly 81% efficient in a worst-case scenario of loading.

It involves bio- inspired spine control, inverse kinematics-based limb control, proper limb-spine coordination, reflex mechanisms and attitude control.

Our results confirm that the design of reflexes like stumbling and extension, combined with an attitude controller, allows for the improving of the performance of the robot in a generic rough terrain which includes stairs, holes and bumps with several levels of complexity adjusted according to the robot dimensions.

In this paper, we present an end-to-end approach for home-environment assistive humanoid robots to provide personalised assistance through a dressing application for users who have upper-body movement limitations.

We use randomised decision forests to estimate the upper-body pose of users captured by a top-view depth camera, and model the movement space of upper-body joints using Gaussian mixture models.

We propose a new aerial tool operation system consisting of multiple quadrotors connected to a tool by spherical joints to perform tool operation tasks.

We model the system and show that the attitude dynamics of each quadrotor is decoupled from the tool dynamics, so that we can consider the quadrotors as thrusters and control the tool by adjusting the orientation and magnitude of these thrusters.

We then design control laws for the tool-tip position/orientation tracking of the (under-actuated) tool system with two quadro- tors and the (fully-actuated) tool system with three quadrotors.

The recorded data allows to characterize the flight behavior of quadrocopters at high airspeeds: As an example, an estimate of the actual thrust produced by the motors and of the aerodynamic drag acting on the vehicle is presented.

Therefore, we herein focus on hand motion and first extract the essential hand motions for knotting, referred to herein as skill motions, by observing human knotting procedures.

We propose a reliable method for estimating the parameters of the terrain quadrupedal robots move on, in the face of limited perception capabilities and drifting robot pose estimates.

By fusing inertial measurements, kinematic data from joint encoders and contact information from force sensors, the local inclination can be robustly estimated and used to optimize the contact forces to reduce slippage.

The estimated terrain information, namely the pitch and roll angles of the ground plane, is exploited in an extended version of our previous model- based control approach.

In this paper, we simulate outdoor disturbances in the lab- oratory setting and investigate the effects of wind gusts on the flight dynamics of a millimeter-scale flapping wing robot.

Each primitive contains visual or force information, a physical model of a sheet of paper is used for analyzing its deformation, and a machine learning method is used for predicting its future state.

We aimed for a function that could rotate the tilt angle continuously and without limit and a function for flying maintaining any desired tilt angle with a structure that could efficiently use the thrust generated by the propellers.

This ability of this mechanism to fly walls with continuously changing surface angles and full 360 spherical coverage makes possible applications in investigation, measurement, etc.

Single joint robotic orthoses for gait rehabilitation: An educational technical review

Robot-assisted physical gait therapy is gaining recognition among the rehabilitation engineering community.

Several robotic orthoses for the treatment of gait impairments have been developed during the last 2 decades, many of which are designed to provide physical therapy to a single joint of the lower limb;

There is a strong need to develop assist-as-needed control algorithms and to perform clinical evaluation of these robotic orthoses in order to establish their therapeutic efficacy.

Epub ahead of print Mar 2, 2016 INTRODUCTION Stroke and spinal cord injury (SCI) are the leading causes of lower limb disability and gait impairment (1).

Patients with these neurological impairments may need to use a wheelchair and be unable to perform activities of daily living (ADL), leading to an increased burden on healthcare and social welfare systems.

The concept of body weight-supported (BWS) physical gait therapy has conventionally been used for the rehabilitation of neurologically impaired patients (5–7).

In the process of BWS gait therapy, the weight of the patient is supported or compensated for and the lower limbs are moved in a repetitive manner by a team of physical therapists in order to restore the patient’s gait functions.

However, it has certain limitations, such as therapist fatigue, reduced number of physical therapy sessions, the non-repetitive nature of training sessions performed by different therapists, and a lack of any objective method to record and analyse the patient’s progress and recovery (11).

Several robotic devices have been developed during the last 2 decades that can provide objective, customized, repetitive and prolonged gait training sessions compared with manual physical gait therapy (16–29).

Most of these robotic devices are wearable exoskeletons, commonly known as “robotic orthoses”, which work in close proximity with the patient’s joints.

Most of the above-mentioned robotic gait training orthoses, such as Lokomat® (Hocoma, Switzerland), are multi-joint devices, which can provide rehabilitation simultaneously to the ankle, knee and hip joints.

Two of these DOF (ankle plantar/dorsiflexion and inversion/eversion) are powered by mechanical actuators, whereas the third, internal/external motion of the ankle joint, is held passive (i.e.

If both motors push or pull in the same direction a plantar/dorsiflexion motion is produced and if they push or pull in opposite directions an inversion/eversion motion is produced.

“Backdrivability” is an important aspect of the design of robotic rehabilitation orthoses and is defined as the extent of freedom provided by the robotic orthosis to the patients to be able to drive the robot voluntarily themselves.

simple, lightweight ankle-foot orthosis (AFO) has been developed at the University of Michigan, Ann Arbor, MI, USA, to provide plantar flexion to the ankle joint during treadmill or over-ground training of neurologically impaired subjects (38–40).

MIT’s ankle-foot orthosis An AFO for stroke patients has also been developed at MIT in order to control the movement of ankle plantar/dorsiflexion in the rehabilitation of drop foot (41).

The bio-inspired AFO has been designed after studying the muscle anatomy of the lower limb and has no rigid frame structure, unlike the above-mentioned devices.

Five different design configurations of exoskeleton are considered to analyse the effects of internal joint forces/torques in the knee joint due to the human-machine interaction.

These design configurations include: pin and fixed end, pin and slider, cam and slider, pin and pinned slider, and cam and pinned slider.

An adaptive knee joint exoskeleton, comprising a pin slider/cam mechanism is designed based on knowledge of knee joint kinematics, which helps in eliminating the negative effects associated with the closed leg-exoskeleton kinematic chain on a human knee (49).

The stiffness control module consists of a friction-based latching mechanism that has been designed to provide 2 levels of stiffness for knee flexion.

The latching mechanism comprises a friction lever, shaft, bearing block, DC motor, worm-gear, cam, spring-loaded push-button and retreat push-button.

A back support with several passive DOF is attached to the user to provide physiological movement to the pelvis and gravity compensation for the device (18, 51).

robotic hip exoskeleton has been designed at the University of Michigan (52) in order to enhance the understanding of biomechanics of human gait as well as providing rehabilitation to neurologically impaired subjects.

Different control algorithms can be developed to provide customized gait rehabilitation according to the disability level and stage of rehabilitation of neurologically impaired subjects (11, 53–55).

Control of robotic gait training orthoses is a rapidly evolving research field and different control algorithms have been designed and evaluated for the above-mentioned robotic orthoses.

Trajectory tracking control Robotic gait training orthoses have traditionally been controlled by simple position control algorithms, commonly known as trajectory tracking or path control.

The trajectory tracking control is useful for the initial phases of rehabilitation when the patients are in bed or using a wheelchair and cannot contribute any effort towards the gait training process.

A model of the human-robotic system has been developed and, based on that model, a path control scheme has been developed (45), which been evaluated for seated positions and has provided the intended results (45).

AAN control schemes have been proposed for single joint robotic orthoses, which estimate the physical capabilities of the patients and modulate the robotic assistance accordingly.

The AAN controller has 3 stages during which it estimates the current phase of gait cycle online, estimates the required assistive torque and transfers the assistive torque to the human subject’s limb (51).

This IM controller monitors the position of the foot continuously throughout the gait cycle and applies the forces necessary for adequate forward motion.

For the controlled plantar flexion, a torsional spring control is applied in order to adjust the stiffness of the AFO joint so that forefoot collisions with the ground are minimized.

A control scheme based on finite-state machine has also been designed for the quasi-passive knee exoskeleton in order to engage the assistance spring (50).

The controller identifies the states of the gait cycle using in-sole sensors that indicate the heel and toe contacts with the ground (50).

A force sensor is used in series with the pneumatic cylinders to regulate the forces applied by the robotic exoskeleton to the hip joint (52).

These robotic orthoses can be divided into full lower limb or multi-joint robotic orthoses (11) and single joint robotic orthoses (37).

Due to this limitation of ankle joint robotic orthoses, physical therapists provide manual training sessions for the DOF that are not provided by the robotic orthoses.

Advance control schemes, such as impedance control, can be utilized to control the compliance of these robotic orthoses actively so that AAN gait training can be provided to neurologically impaired subjects.

Such control schemes have already been proposed for the multi-joint robotic gait training orthoses powered by PMA (19, 20) and can be adapted to the single joint orthoses.

The alignment of these robotic orthoses, especially, hip and ankle joint orthoses with anatomical joints, present a major design challenge because the hip and ankle are complex anatomical joints.

An attempt has been made to design a knee robotic orthoses based on the anatomical joint features, but no such attempt has been reported for ankle and hip joints (49).

Bio-inspired Knee Joint Mechanism for a Hydraulic Quadruped Robot (ICARA'15)

Bio-inspired Knee Joint Mechanism for a Hydraulic Quadruped Robot Hamza Khan, Roy Featherstone, Darwin G. Caldwell, Claudio Semini International Conference on Automation, Robotics and Applications...

Improvement of Humanoid Walking control by compensating Actuator Elasticity: experimental result

The actuators in humanoid robots inevitably have compliance in their joint mechanisms. The joint elasticity often negatively affects static and dynamic performance of the robot. In the specific...

Multifilament Musculoskeletal Robot - Behold The Future

Behold The Future...Suzumori Endo Lab, Tokyo Institute of Technology has developed the aritificial neck of musculoskeletal robot driven by multifilament artificial muscles. The main structures...

Magneto Rheological Linear Clutch for Force Controlled Human Safe Robots

This paper proposes a Magneto Rheological linear clutch for use in human safe robotic applications. The force transmitted to the links of the robot must be precisely controlled for any manipulator...

Three-Finger Grasping and Manipulation

A semester project of UM-SJTU JI course VM467 (Intro to Robotics) Three fingers grasp and manipulate a ball. Simulated results are based on the paper by S. Ryew and H. Choi, "Double Active...

Flexible Muscle-Based Locomotion for Bipedal Creatures

We present a control method for simulated bipeds, in which natural gaits are discovered through optimization. No motion capture or key frame animation was used in any of the results. For more...

BB8 finally completed !!!

after a year I have finally completed bb8... not an easy project but the results are great.. enjoy !

Dynamic Torque Control of a Hydraulic Quadruped Robot (ICRA'12)

Dynamic Torque Control of a Hydraulic Quadruped Robot Thiago Boaventura, Claudio Semini, Jonas Buchli, Marco Frigerio, Michele Focchi, Darwin G. Caldwell IEEE International Conference in...

Path Planning with Force-based Foothold Adaptation for Torque Controlled Quadruped Robots (ICRA'14)

Path Planning with Force-based Foothold Adaptation for Torque Controlled Quadruped Robots Alexander Winkler, Ioannis Havoutis, Stephane Bazeille, Jesus Ortiz, Michele Focchi, Darwin G. Caldwell,...

SMA project: flower petal

This flower petal was made by using the curling mechanism (sewing flexinol wire to paper). It was made with 0.006" HT flexinol, which requires about 0.400A of current to run properly. This...