Primary goal of walking is transferring the center of mass (CoM) between initial and target positions. While there exist many different forms of movement in nature only humans have been able to develop bipedal gait in their evolution. From biomechanical point of view bipedal gait is considerably demanding. Large portion of body mass is located high above the walking surface [
1], and is supported by two legs that constitute relatively small base of support. Mastering proper coordination of such demanding locomotion apparatus that ensures good balance during walking also in presence of unexpected disturbances (e.g. slip, misstep, push...) requires good cooperation between biomechanical and neurophysiological processes.
However biomechanical and neurophysiological processes that underlay well balanced bipedal walking are complex and relatively unknown. The lack of deeper understanding of balance mechanisms may limit the potential of neurorehabilitation. Neurological injury of motor cortex (e.g. after stroke) often significantly hinders motor functions which inevitably causes balance problems. Nevertheless human brain can to some extent recover by bypassing damaged area and by re-establishing neural connections in neighbouring areas (neuroplasticity) [
2‐
4] and in this way set up the conditions that provide the potential to restore gait function. Since the human brain is most susceptible to such recovery early after neurological injury, rehabilitation success depends on early start and patient specific therapeutic program. There is a strong initiative in the field of neurorehabilitation to develop novel training paradigms that would appropriately address also balance of bipedal gait.
The most common approach to studying balance is by applying unexpected perturbations to humans while they walk to cause kinematic and dynamic deviations that central nervous system addresses with appropriate motor responses [
5‐
10]. Depending on the body segment that the perturbation acts upon we differentiate between distal and proximal perturbations. Distal perturbation applied base-of-support displacement at feet by suddenly moving the platform under stance leg [
5,
6,
11] in medio/lateral or anterio/posterior direction to mimic slip [
7,
12‐
15] in frontal or sagittal plane respectively. Alternatively subjects may be exposed to sudden obstacles in the line of walking to mimic tripping [
8‐
10,
16] or they need to negotiate sinking platform to mimic misstep [
17]. These experiments are almost exclusively conducted during overground walking. Primary concern of these paradigms is investigation of compensatory postural responses after inducing trips or slips as slips and trips are leading cause of falls and associated injuries in older adults. However if exposed to reoccurring slipping or tripping situations younger and older adults can improve compensatory responses associated with frequency as well as number and length of compensatory steps and time needed to stabilize after perturbation and reduces the incidence of falls [
12‐
16]. In contrast, gait rehabilitation and balance recovery in neurologically impaired subjects do not target specific situations such as trip or slip (although trip and slip are great concerns in any impaired gait) but concern primarily with fundamental mechanisms associated with walking like forward propulsion, weight transfer, cyclical leg movement and dynamic balance that deteriorated due to changed muscle control after neurological damage. One particular concern associated with simulating slip is that moving platforms that are used to apply slipping perturbation [
5,
6] actually deliver two successive perturbations, first one originating from platform acceleration and second one associated with platform deceleration which may be difficult to interpret as both elicit compound responses. For this reason single force proximal perturbations at pelvis are in gait rehabilitation after neurological damage more suitable as they evoke more elementary postural mechanisms as opposed to complex compound mechanisms. Only few studies have investigated dynamic balance responses originating from proximal perturbations in frontal plane in neurologically healthy population walking on a treadmill. Authors have shown that the majority of balance activities after lateral perturbation may be attributed to “stepping strategy” which depending on perturbation direction can manifest as “inward strategy” or “outward strategy” [
18,
19]. Perturbation apparatus was rather simple pneumatically activated single degree of freedom (DoF) mechanism which may not be suitable for accessing dynamic balancing responses also in neurologically impaired population. There is a strong need for multi-DoF pelvis manipulation mechanism that would be able to follow natural pelvis movement during walking and that could also deliver well defined proximal perturbations in the transversal plane. Additionally, such pelvis robot would also need to be able to provide assistive force field to the pelvis to provide adequate support in walking of neurologically impaired population. Different approaches have been proposed to address these needs. Pneumatic actuation was used in [
20] where three pneumatically driven degrees of freedom (DoF) were combined with two passive DoF providing five DoF in pelvis (three translations and two rotations). Linear electromagnetic actuation was implemented in the system [
21] that had the capacity to deliver active force field in horizontal plane. System designed in [
22] conceived lightweight and modular design composed of remote motors, pulleys, force sensors and cables that connect to selected attachment points on the hip brace to interact with pelvis movement. All three solutions were developed for use in combination with treadmill. Another important aspect when addressing dynamic balance during walking relates to walking conditions. So far studies that investigated balance mechanisms in relation to proximal perturbations during walking were conducted while walking on treadmill which may differ from balancing during overground walking.
The goal of our work was to develop a balance assessment robot (BAR) that could be used for studying balance responses during overground walking and would be able to deliver desired force field to the pelvis of a walking subject. In this paper we first present mechanics and control approach of BAR and explore their characteristics. Secondly, BAR was used to assess normative balance responses in a group of neurologically healthy individuals that may serve as a reference in subsequent studies in neurologically impaired population.