Posted on

Reactive Velocity Control Reduces Energetic Cost of Jumping with a

Robots capable of dynamic locomotion behaviors and high-bandwidth sensing with their limbs have a high cost of transport, especially when locomoting over highly dissipative substrates such as sand. We formulate the problem of reducing the energetic cost of locomotion by a Minitaur robot on sand, reacting to robot state variables in the inertial world frame without modeling the ground online. Using a bulk-behavior model of high-velocity intrusions into dry granular media, we simulated single jumps by a one-legged hopper using a Raibert-style compression-extension virtual leg spring. We compose this controller with a controller that added damping to the leg spring in proportion to the intrusion velocity of the robot’s foot into the simulated sand while the robot is pushing off in the second half of stance. This has the effect of both reducing the torque exerted by the motors because the added virtual “active damping” force acts in opposition to the virtual leg spring force, and reducing the transfer of energy from the robot to the sand by slowing the intrusion velocity of the foot. Varying the simulated robot’s initial conditions and the simulated ground parameters, we gained a consistent 20% energy savings by adding active damping with no cost in apex height.

For more information, see the Kod*lab website: kodlab.seas.upenn.edu

Posted on

Actuator Transparency and the Energetic Cost of Proprioception

In the field of haptics, conditions for mechanical “transparency”[1] entail such qualities as “solid virtual objects must feel stiff” and “free space must feel free”[2], suggesting that a suitable actuator is able both to do work and readily have work done on it. In this context, seeking actuator transparency has come to mean a preference for minimal dynamics [3] or no impedance [4]. While such general notions seem satisfactory for a haptic interface, actuators with good mechanical transparency are now being used in high-performance robots [5, 6] where once again they must be able to do work, but are now also expected to perceive their environment by processing signals related to contact forces in the leg or manipulator when an explicit force sensor is not present. As robotics researchers develop models [7] suitable for programming behaviors that require systematic making and breaking of contact within the environments on which they perform work, actuators must be capable of: (a) generating the high forces at speed needed to accelerate the body during locomotion [5]; (b) robustness to high forces and impacts during locomotion [8]; (c) perceiving high force events quickly, such as touchdown in stance [9]; (d) perceiving contact quickly without exerting significant force on the object, such as in gentle manipulation [10]; and (e) reacting quickly during time-sensitive behaviors [11].

This work aims to describe a quantitative assay of transparency that might, for example, predict the advantage in proprioceptive tasks of an electromagnetic directdrive (DD) motor (i.e., one without gearbox), relative to actuation schemes consisting of both a motor and a geared reduction. Specifically, we explore the prospects for characterizing transparency as revealed by comparing the energetic cost of “feeling” the environment. Our sample proprioceptive task is instantiated by a simple torque estimator in Sec. 2. This scheme is then instrumented in simple contact detection experiments paired with a model to empirically explore the relationships between collision energy and detection time delay in Sec. 3. The actuators are then tested with a feel-cage task to illustrate the advantage of good transparency in Sec. 4.

“For more information: Kod*lab (link to kodlab.seas.upenn.edu)

Posted on

Sensor-Based Reactive Execution of Symbolic Rearrangement Plans by a Legged

We demonstrate the physical rearrangement of wheeled stools in a moderately cluttered indoor environment by a quadrupedal robot that autonomously achieves a user’s desired configuration. The robot’s behaviors are planned and executed by a three layer hierarchical architecture consisting of: an offline symbolic task and motion planner; a reactive layer that tracks the reference output of the deliberative layer and avoids unanticipated obstacles sensed online; and a gait layer that realizes the abstract unicycle commands from the reactive module through appropriately coordinated joint level torque feedback loops. This work also extends prior formal results about the reactive layer to a broad class of nonconvex obstacles. Our design is verified both by formal proofs as well as empirical demonstration of various assembly tasks.

For more information: Kod*lab

Posted on

Vertical hopper compositions for preflexive and feedback-stabilized quadrupedal bounding, pacing,

This paper applies an extension of classical averaging methods to hybrid dynamical systems, thereby achieving formally specified, physically effective and robust instances of all virtual bipedal gaits on a quadrupedal robot. Gait specification takes the form of a three parameter family of coupling rules mathematically shown to stabilize limit cycles in a low degree of freedom template: an abstracted pair of vertical hoppers whose relative phase locking encodes the desired physical leg patterns. These coupling rules produce the desired gaits when appropriately applied to the physical robot. The formal analysis reveals a distinct set of morphological regimes determined by the distribution of the body’s inertia within which particular phase relationships are naturally locked with no need for feedback stabilization (or, if undesired, must be countermanded by the appropriate feedback), and these regimes are shown empirically to analogously govern the physical machine as well. In addition to the mathematical stability analysis and data from physical experiments we summarize a number of extensive numerical studies that explore the relationship between the simple template and its more complicated anchoring body models.

For more information: Kod*lab

Posted on

Analytically-Guided Design of a Tailed Bipedal Hopping Robot

We present the first fully spatial hopping gait of a 12 DoF tailed biped driven by only 4 actuators. The control of this physical machine is built up from parallel compositions of controllers for progressively higher DoF extensions of a simple 2 DoF, 1 actuator template. These template dynamics are still not themselves integrable, but a new hybrid averaging analysis yields a conjectured closed form representation of the approximate hopping limit cycle as a function of its physical and control parameters. The resulting insight into the role of the machine’s kinematic and dynamical design choices affords a redesign leading to the newly achieved behavior.

Posted on

Sensor-Based Reactive Symbolic Planning in Partially Known Environments

This paper considers the problem of completing assemblies of passive objects in nonconvex environments, cluttered with convex obstacles of unknown position, shape and size that satisfy a specific separation assumption. A differential drive robot equipped with a gripper and a LIDAR sensor, capable of perceiving its environment only locally, is used to position the passive objects in a desired configuration. The method combines the virtues of a deliberative planner generating high-level, symbolic commands, with the formal guarantees of convergence and obstacle avoidance of a reactive planner that requires little onboard computation and is used online. The validity of the proposed method is verified both with formal proofs and numerical simulations.

For more information: Kod*lab

Posted on

On Balancing Event and Area Coverage in Mobile Sensor Networks

In practice, the mobile sensor networks have two important tasks: firstly, sensors should be able to locate themselves close to where major events are happening so that event tracking becomes possible; secondly, the sensor networks should also maintain a good area coverage over the environment in order to detect new events. Because these two tasks are usually conflicting with each other, a coverage control policy should be able to balance the event and area coverage of the environment. However, most existing work is to achieve either optimal event coverage or optimal area coverage over the environment. In this thesis, a Voronoi-based coverage control with task assignment is introduced: each sensor is allowed to switch between event and area coverage depending on the intensity of events within its Voronoi cell, and both continuous-time and discrete-time control for sensor positions are discussed.

Posted on

Nomadic Monument for Women in Robotics

We describe the Nomadic Monument for Women in Robotis (NMWR), a project celebrating women pioneers in robotics. NMWR is a 13’ semi-transparent geodesic dome with illustrations and descriptions of the women and their research on the inside faces of the triangles. Visitors can see rough outlines of the illustrations from the outside, but must enter the dome to learn about the women. As an immersive environment, the dome provides visitors not only a place to learn about inspiring women but also a space to feel that they are a member of this inspiring community.

The geodesic dome was introduced by Buckminster Fuller in the mid-20th century as a new form for human shelter that would be more environmentally friendly and affordable, made possible by new building materials and the innovative use of tension to hold a built structure together. Simple facts and diagrams about dome geometry are included on dome. Fuller’s domes have also had a lasting influence on artists, including Krieger. Her 11-year ongoing project Plastic Fantastic engages community members in building a geodesic dome out of 6000 post-consumer water bottles, with notable installations at the Anchorage Museum in 2014 and Philadelphia’s inaugural Art in the Open in 2010. The NMWR dome will be constructed out of PVC piping and light blue, red, and purple panels of acrylic.

A range of living women roboticists, many with a connection to Philadelphia, are featured in the dome. They work on a variety of problems in modern robotics, including helping give robots a sense of touch (Katherine Kuchenbecker, formerly at Penn); programming groups of robots to work together (Daniela Rus, MIT); sensing devices with medical applications (Ruzena Bajcsky, founder of the GRASP Lab at Penn); designing robots to move around in and collect information about harsh environments like Mars and Antarctica (Ayanna Howard, Georgia Tech); developing technologies (including robots) to assist in caring for children or the elderly (Maja Mataric, University of Southern California), and many more.

The portraits and descriptions of these women challenge stereotypes about who can be an engineer, and the presentation of Hartmann-Dow’s near life-sized portraits at eye level around the inside of the dome creates a space for visitors to enter the community of women roboticists. Stereotype challenge and the sense of belonging to a community have both been shown to influence interest in engineering by women and girls (for a review, see [1]).

NMWR will be presented at the 2018 Philadelphia Science Festival and will be on view at three events throughout the festival: At the Screening of Top Secret Rosies on Friday, April 20; at the Be a Pennovator event on Sunday, April 22; and at the Science Carnival on Saturday, April 28. The first two events will be at the Pennovation Center, and the last will be on the Parkway.

[1] Cheryan, Sapna, Sianna A. Ziegler, Amanda K. Montoya, and Lily Jiang. “Why Are Some STEM Fields More Gender Balanced than Others?” Psychological Bulletin 143, no. 1 (2017): 1.

For more information: Kod*lab.

Posted on

A hybrid dynamical extension of averaging and its application to

We extend a smooth dynamical systems averaging technique to a class of hybrid systems with a limit cycle that is particularly relevant to the synthesis of stable legged gaits. After introducing a definition of hybrid averageability sufficient to recover the classical result, we illustrate its applicability by analysis of first a one-legged and then a two-legged hopping model. These abstract systems prepare the ground for the analysis of a significantly more complicated two legged model—a new template for quadrupedal running to be analyzed and implemented on a physical robot in a companion paper. We conclude with some rather more speculative remarks concerning the prospects for further extension and generalization of these ideas.

Posted on

Integration of Local Geometry and Metric Information in Sampling-Based Motion

The efficiency of sampling-based motion planning algorithms is dependent on how well a steering procedure is capable of capturing both system dynamics and configuration space geometry to connect sample configurations. This paper considers how metrics describing local system dynamics may be combined with convex subsets of the free space to describe the local behavior of a steering function for sampling-based planners. Subsequently, a framework for using these subsets to extend the steering procedure to incorporate this information is introduced. To demonstrate our framework, three specific metrics are considered: the LQR cost-to-go function, a Gram matrix derived from system linearization, and the Mahalanobis distance of a linear-Gaussian system. Finally, numerical tests are conducted for a second-order linear system, a kinematic unicycle, and a linear-Gaussian system to demonstrate that our framework increases the connectivity of sampling-based planners and allows them to better explore the free space.

For more information: Kod*lab.