Humans are, by nature social entities with a remarkable capacity to learn a variety of motor skills, ranging from catching a ball at a young age to jumping and landing on a trampoline. The ability to touch and feel can be described by the haptic sense, which is an integral part of our everyday experience that few of us really notice. Even a small, silent vibrational signal of a smartphone, perceived by the user leads to the acknowledging reactions, such as taking the phone and answering the received message with a call or a text message. Within the scope of this work, the main task is to couple a mathematically expressed running model with a haptic interface. Especially in the field of sensimotor learning, human-in-the-loop haptic modelling of locomotion is a recent subject. The main idea is to apply results – gained from this work – to control a robotics interface paradigm for exoskeletons (B.A.L.A.N.C.E, 2013). Since the control of exoskeleton involves continuous interaction with human body, only computational modelling is not sufficient, as it is hard to model human muscle-skeletal activity. This work deals with human- machine interaction using a 1-degree of freedom haptic device controller. Based on the human-in-the-loop approach (HITL), a haptic interface is deployed, which provides mechanical input forces to compute and model the locomotion of running.

The approaches shown in this thesis are based on human motor control skills with the aim to learn and stabilize a virtual running model under unstable conditions. Basically, there are many options on how to introduce parameter changes to the SLIP running model. In the following, two cases are taken into consideration:

• Applying an external force to the point mass in real-time
• Changing the spring stiffness in real-time

It should be noted, that all forces and parameter changes are applied during stance phase. More information concerning different approaches for stabilizing locomotion patterns can be retrieved in section 8 (Future Work). By doing experiments on two (different) cases, a research hypothesis is formulated stating that:

“Spring stiffness manipulations in real-time are more effective than applying an external force to the point mass with respect to achieve a stable running pattern.

In order to answer this question, first thoughts on the topic of general locomotion must be made. Locomotion is an interdisciplinary field of studies, e.g. the generation of movement starts in the human brain, where somatosensory information is flowing through the nerves received by the motor neurons located in the muscle. Taking a common example, e.g. trying to catch the train which is departing in 5 minutes and the distance is still too large for walking. In order to get to the train in time, velocity needs to be incremented and the movement pattern changes from walking to running. What looks so easy in practise, can be very complex when it comes to a mathematical description. Also the complex task of synergetic activation accomplished by the neuromuscular system seems to be easy in practise. In order to understand the basic mechanism of human movement strategies, a simplified model needs to be implemented, which is taking the main forces acting on the human body into consideration.

Haptic interaction with the human being can be regarded as a milestone in modern technology of prosthetics and robotics. The previously mentioned (running) model needs to be coupled with a haptic device interface, which has been developed at Imperial College (2017). This interface is an input-output hardware module, such e.g. a joystick, which can be controlled by the user and also provide feedback. In order to understand the basic functionalities of haptics, it is relevant to know the main technical specifications about the haptic control interface which will be explained into detail in section 3.3 (System Description). The main user task is to control this haptic device, i.e. to manipulate a certain parameter of the hardware-coupled spring-loaded inverted pendulum (SLIP) model in real-time. Either, the stiffness parameter can be manipulated by the haptic device robot, or an additional force is applied externally to the point mass. Parameter manipulation of the implemented SLIP model is updated in real-time by the human user. Thus, the design should be kept as simple as possible in order to make it usable for everyone. This approach is called human-in-the-loop (HITL). In this approach, the human user is the man/manipulator in the middle. Multiple feedback signals (visual and tactile force feedback; see 3.2 Role of Feedback) are fed as an input to the user. Based on the feeding input, the user is able to manipulate parameters in real-time, haptically. The HITL-approach forms an endless loop in case that the SLIP model doesn’t fall down. Sensimotor training may lead to task improvement, when it comes to control of the haptic device robot. Since every user is unique, there are multiple solutions for different parameter configurations. When it comes to sensimotor learning tasks, it is important, that a goal must be specified. How to achieve this goal is up to the haptic control skill level of each user.

indexmenu_n_10