Kapandji Thumb Opposition Test Overview

• Kapandji Thumb Opposition Test is a standardized protocol that assesses thumb opposability by having the thumb touch 10 predefined positions on the hand.
• The test uses a binary scoring system where each successful contact is marked, providing measurable insight into mechanical dexterity.
• In both clinical and robotic contexts, the test informs design efficiency and control strategies by highlighting key performance and design limitations.

The Kapandji Thumb Opposition Test is a standardized clinical and experimental protocol that quantitatively assesses thumb opposability by evaluating a hand’s capacity to position the thumb tip at a series of anatomically significant locations across the palm and fingers. Originally described by Kapandji (1986), this test is widely adopted in both human hand function assessment and in the evaluation of robotic and prosthetic hands seeking to mimic human dexterity. In robotics, the test provides a reproducible score for thumb workspace and opposability, directly reflecting the mechanical design and actuation strategies of an artificial hand.

1. Definition and Protocol of the Kapandji Thumb Opposition Test

The Kapandji test consists of commanding the thumb to touch a sequence of pre-defined locations (“Kapandji positions”) along the palmar surface of the same hand. The canonical protocol specifies 10 or 11 discrete positions, progressing from the radial (index-finger side) to the ulnar (little-finger side) border. The most commonly accepted positions are as follows:

Kapandji Position	Anatomical Target	Index
1 (or 0)	Radial base of index-finger metacarpal/base of index	0/1
2	Radial side of index PIP region	2
3	Pulp of index fingertip	3
4	Pulp of middle fingertip	4
5	Pulp of ring fingertip	5
6	Pulp of little fingertip	6
7	Proximal phalanx of little finger	7
8	Middle phalanx of little finger	8
9	Hypothenar eminence (proximal palm)	9
10	Distal palmar crease or volar wrist border	10

A pass/fail determination is made for each position: a “success” ($s_i = 1$) is scored if the thumb tip establishes visually confirmed contact with the target; otherwise $s_i = 0$. The final Kapandji score is the sum over all $N$ positions: $\text{Score} = \sum_{i=1}^{N} s_i$. This protocol is used in both clinical settings and in robotic hand research (Puhlmann et al., 2022, Weng, 20 Apr 2025).

2. Robotic Implementation and Evaluation Methodology

Robotic hands are typically evaluated using the Kapandji test by commanding the thumb through a series of pre-defined poses, each associated with a target contact. The protocol in leading robotic studies exhibits the following characteristics:

• Actuation Control: Hands such as RBO Hand 3 utilize pre-recorded pneumatic actuation profiles, while cable-driven platforms like BiDexHand employ joint-level position control mapped onto servomechanisms driving rotational degrees of freedom.
• Contact Determination: In both RBO Hand 3 (Puhlmann et al., 2022) and BiDexHand (Weng, 20 Apr 2025), no force or pressure sensors are used to verify contact. Instead, contact is confirmed by direct visual inspection, consistent with clinical implementations.
• Scoring and Metrics: A binary scoring per position defines the outcome; no explicit measurement of contact force, timing, angular error, or repeatability is reported. There are no distance thresholds, torque measurements, or analytic formulas for score computation.
• Actuated Components: For RBO Hand 3, four thumb degrees of actuation (DOA) plus an actuated palm are used. BiDexHand deploys four actively actuated thumb joints with passive linkage-driven interphalangeal joint (IP) movement.
• Palm and Whole-Hand Contribution: In hands that incorporate palm actuation (e.g., RBO Hand 3’s “palm bellow”), certain thumb opposition poses, especially in the ulnar region, can only be achieved when the palm is actively hollowed or rotated.

3. Representative Robotic Hand Results

Two recent anthropomorphic robotic hands provide state-of-the-art reference implementations:

• RBO Hand 3 (Puhlmann et al., 2022):
  • Achieves all ten Kapandji positions, yielding a perfect score of 10/10.
  • Uses 16 DOA (including four for the soft pneumatic thumb and one for the palm).
  • No per-position timing or force measurement is reported.
  • Demonstrates that palm actuation is essential for achieving ulnar positions (5–10).
  • Intrinsic compliance and passive shape adaptation enable successful open-loop execution.
• BiDexHand (Weng, 20 Apr 2025):
  • Achieves contact for 9 out of 11 positions (using 0–10 labeling): fails at metacarpal root of index (#0) and middle phalanx of little finger (#8), directly due to limitations in carpometacarpal (CMC) range.
  • Employs four independent thumb actuators and passive anti-parallelogram coupling for the thumb IP.
  • Implements CMC abduction/adduction and flexion/extension as decoupled revolutes, distinguishing from the human saddle joint.
  • No force sensing or closed-loop feedback at the fingertip.
  • Cable routing friction and lack of compound CMC motion at the extremes are primary limiting factors.

Both platforms use visually judged pass/fail scoring to determine test outcomes and employ joint-level control tailored to their mechanical architectures.

4. Mechanical and Control Design Considerations

The results of the Kapandji test for a robotic hand are direct reflections of mechanical design choices:

• Thumb Kinematics:
  • The presence of multiple actively controlled thumb DOAs, including abduction/adduction and flexion/extension at CMC and MCP, is critical for opposability.
  • In BiDexHand, the use of an anti-parallelogram four-bar linkage for the thumb and finger IP joints reduces actuator count while preserving coordinated distal flexion.
• CMC Joint Implementation:
  • Human thumb CMC is a saddle joint coupling flexion and abduction; most robotic implementations (e.g., BiDexHand) use uncoupled revolutes, reducing the compound range and limiting opposition at extremes.
  • RBO Hand 3’s soft pneumatic thumb, leveraging compliance and modular actuators, achieves workspace sufficient for the highest Kapandji score.
• Palm and Finger Contribution:
  • Active palm actuation, as in the RBO Hand 3, enables positioning of the ring and little finger targets within reachable workspace for the thumb.
  • Mechanisms without palm actuation or with limited finger ab/adduction (e.g., prior RBO Hand 2) cannot achieve full opposition scores.

A comparison of mechanical strategies for selected hands is summarized:

Hand	Thumb DOA	Palm DOA	CMC Type	Kapandji Score
RBO Hand 3	4 (soft)	1 (bellow)	Soft (uncoupled)	10/10
BiDexHand	4 (servo + linkage)	0	Decoupled revolute	9/11
BCL-26 (Zhou)	>4	Yes	Not specified	10/10

5. Limitations and Prospective Design Improvements

Robotic failures to achieve higher Kapandji opposition scores are typically caused by mechanical constraints or decoupling of thumb axes at the CMC joint. Specific limitations include:

• In BiDexHand, inability to reach certain palmar targets results from restricted compound CMC flexion/abduction due to serial, rather than coupled, implementation.
• Cable-drive mechanisms introduce friction and backlash, particularly for small compound motions at kinematic extremes, reducing fine positional accuracy.
• Open-loop position control without feedback does not compensate for compliance, friction, or joint hysteresis.
• The lack of closed-loop force or tactile sensing precludes dynamic compensation during the opposition maneuver.

Recommendations for future improvements derived from the evaluated platforms include:

• Implementing a biomimetic CMC saddle joint to achieve true compound movements.
• Incorporation of closed-loop tactile or visual feedback to adaptively control the thumb trajectory and ensure successful contact.
• Optimizing tendon routing and reducing friction to minimize backlash in cable-driven architectures.
• Exploring synergy-based couplings or adding extension actuators to increase thumb postural stability at extreme opposition.

A plausible implication is that with more biomimetic joint architectures and the incorporation of sensor feedback, future hands may routinely achieve or surpass the full Kapandji range attained by current models (Weng, 20 Apr 2025).

6. Significance in Robotic and Prosthetic Hand Research

The Kapandji test is recognized as a critical benchmark in the robotics community for the evaluation of anthropomorphic hand designs. Attainment of a high Kapandji score signifies sufficient mechanical opposability for dexterous manipulation tasks, paralleling clinical assessments of human hand rehabilitation. It provides a direct, reproducible, and interpretable index of mechanical design effectiveness, actuation sufficiency, and system-level integration.

Robotic hands achieving the maximum Kapandji score (e.g., RBO Hand 3 (Puhlmann et al., 2022), Zhou’s BCL-26) are validated in their opposable workspace as capable of a breadth of grasp types and manipulation strategies prescribed by the GRASP taxonomy. Lower scores indicate specific, localizable mechanical or control deficiencies, informing targeted redesign of thumb kinematics or palm architecture.

In summary, the Kapandji Thumb Opposition Test remains an indispensable metric for rigorous, quantitative comparison of robotic hand architectures, bridging clinical and engineering domains and providing actionable diagnostic granularity for future developments in dexterous manipulation.